Identification of Hazardous Areas for Priority Landmine Clearance: AI for Humanitarian Mine Action

Identification of Hazardous Areas for Priority Landmine Clearance: AI for Humanitarian Mine Action

TL;DR: Landmines pose a persistent threat and hinder development in over 70 war-affected countries. Humanitarian demining aims to clear contaminated areas, but progress is slow: at the current pace, it will take 1,100 years to fully demine the planet. In close collaboration with the UN and local NGOs, we co-develop an interpretable predictive tool for landmine contamination to identify hazardous clusters under geographic and budget constraints, experimentally reducing false alarms and clearance time by half. The system is being tested in Afghanistan and Colombia, where it has already led to the discovery of new landmines.


Anti-personnel landmines are explosive devices hidden in the ground designed to explode by proximity or contact and with the capacity to kill, disable or cause harm to humans (Fig. 1). The mere threat of landmine contamination in a territory not only endangers the physical well-being of affected populations but also results in a loss of forest areas, reduction of productive land, exacerbation of social vulnerability, delay of infrastructure development, and damage of natural, physical, and social capital. Due to such negative consequences, in 1997 most countries signed the Ottawa Treaty committing themselves to stop the manufacture, commercialization, and use of landmines. Likewise, the countries that had historically used these explosive devices during armed conflicts undertook to clear the contaminated territories. Despite ongoing efforts, landmines continue to be used in conflicts worldwide, posing a persistent threat to humanity and hindering the development of war-affected communities in over 70 countries, impacting more than 60 million people and causing nearly 7,000 casualties every year.

Figure 1. Example of a landmine found in Colombia.

Humanitarian mine action operations seek to clear conflict-affected regions of remaining landmines so that communities can safely reland their territories. However, demining operations are laborious and costly due to vast areas that need surveying and the limited monetary and human resources available: at the current rate, it will take about 1,100 years to clear the planet of all remaining landmines, underscoring the urgent need for innovative evidence-based approaches to make demining operations more efficient and safer. In this context, we co-designed the RELand system (Risk Estimation of Landmines), in partnership with the United Nations Mine Action Service and local demining organizations, to efficiently identify hazardous areas for priority landmine clearance. RELand is currently being tested in Colombia, where it has already led to the discovery of three new landmines in a newly prioritized area, potentially saving civilian lives. We have also tailored and deployed the system in Afghanistan, and we are preparing for its deployment in war-torn territories globally, in partnership with UNMAS and UNOPS.

RELand: Risk Estimation of Landmines via Interpretable Invariant Risk Minimization

RELand is a holistic pipeline to identify priority hazard areas to support non-technical surveys in humanitarian demining operations. Theses initial surveys are currently carried out by human experts who evaluate the possible presence of landmines based on available information and that provided by the residents. Since landmines are not used randomly but under war logic, Machine Learning can potentially help with these surveys by analyzing historical events and their correlation to relevant features. However, identifying landmine contamination has been scarcely studied in the literature, and poses three main challenges: noisy labels, geographic dependence, and sparse predicted risk scores. We address the challenges of landmine risk estimation by enhancing existing datasets with rich relevant features, constructing a novel, robust, and interpretable ML model that outperforms standard and new baselines, and identifying cohesive hazard clusters under geographic and budgetary constraints. Finally, the results are delivered through a web application developed with key mine action stakeholders. The major components of RELand are illustrated in Fig. 2. Notably, our approach is the first public pipeline of its kind that can be easily adapted for use in demining workflows globally.

Figure 2. Integration of RELand system into the humanitarian demining pipeline. Current non-technical surveys (grey) are based on the visual inspection of data in geospatial information systems and human expert analyses including local community surveys and domain knowledge. RELand (yellow dashed box) serves as an additional toolbox that contains three major components: dataset enhancement based on existing public geospatial datasets (red), risk modeling with machine learning methods (blue), and interactive web interface (green).

The first component of the system, Dataset Enhancement, integrates different sources of information to construct a dataset for landmine presence with rich relevant features based on geographic information, socio-demographic variables, remnants of war indicators, and historical landmine events. We introduce several new features which prove useful to identify hazard areas and to rule out false alarms. We also argue how labels should be assigned to predict the results of humanitarian demining operations, rectifying the definition of labels used in previous literature.

For the Risk Modeling component, we designed a novel interpretable deep learning tabular model extending TabNet. We propose to minimize the Invariant Risk Minimization (IRM), which enables the model to be robust to distribution shifts and invariant to diverse deployment environments. Intuitively, we define an “easy” environment as one where landmines are found close to past events or grid cells with no historical landmines nearby have indeed negative labels. In contrast, a “hard” environment is one where despite there being some historical events there are no new landmines (and resources are going to be used inefficiently) or new landmines found far away from previous events (and likely missed by baseline methods leading to a latent risk to humans). Formally, let us denote an environment by (e = (X^e, Y^e)) and let (w) be a dummy scalar classifier. Then the IRM loss is composed by an ERM cross-entropy term that encourages prediction accuracy, and a regularization term that forces (f_theta) to be simultaneously optimal across all environments (E). Our landmine risk estimator (f_{theta}(X)) is penalized for applying the distance-existence rule in “easy” environments to “hard” ones, and therefore generalizes well on both environments.

$$IRM(theta) = min_{theta} sumlimits_{e in E} ell_{text{CE}}(f_theta(X^e), Y^e) + lambda cdot ||nabla_{w|w=1} ell_{text{CE}}(w( f_theta(X^e)), Y^e)||^2$$

However, our partner demining organizations quickly emphasized the need for interpretable models, as they must explain to communities why certain areas are prioritized for clearance or not. Therefore, as the first step towards the interpretation of landmine risk estimators, we utilize SparseMax layers to generate global feature importance for our model. SparseMax (SM) is an activation function that normalizes the input vector to sparse probabilities (like a LASSO regularization), and is shown at the top of Fig. 3. Finally, we leverage the sequential design in TabNet to form decision blocks that are summed together and passed into an aggregation FC layer as the final prediction. This sequential design resembles additive modeling in Gradient Boosting Machines and ResNet skip connection mechanism. Initial blocks capture the main correlation in the dataset, and the following blocks can use the rest of the features to learn the residuals to fit the function better. Our final architecthure is show in Figure 3.

Figure 3. RELand architecture with interpretation branch that generates sparse feature masks on the top, and decision blocks at the bottom aggregated before the final FC layer.

To validate the proposed system, we simulate different scenarios in which the RELand system could be deployed in mine clearance operations using real data from Colombia. We use a block cross-validation approach, where the hold-out set corresponds to all cells in a municipality, to account for the geographical nature of current demining operations. In addition, since false negatives represent a higher cost in terms of human lives, we use the Height and Reverse Height (rHeight) metrics of how well a ranking is generated, in the sense that positive cells should be ranked higher than negative cells. Intuitively, models with better predictions for top-ranked regions can speed-up land clearing operations. Given a predicted risk score, Height refers to the number of positive cells ranked below a negative cell, and rHeight is the number of negative cells ranked above a positive one. An ideal classifier minimizes both of these metrics and perfectly rank positive cells above negative cells. Formally,

$$ Height(X_n) = sumlimits_{i = 1}^{P}mathbb{1}(widehat{f}(X_text{p})_i leq widehat{f}(X_text{n})), $$

$$rHeight(X_p) = sumlimits_{j = 1}^{N}mathbb{1}(widehat{f}(X_text{p}) leq widehat{f}(X_text{n})_j) $$

where (P) and (N) are the total counts of positive and negative labels, respectively, and (widehat{f}(X_text{p})) ((widehat{f}(X_text{n}))) is the predicted probability when the ground truth of (X_i) ((X_j)) is positive (negative).

Table 1 presents the result of the experimental validation comparing the proposed methodology with current practices, focusing mainly on historical landmine reports, and two previous ML models proposed in the literature. RELand consistently outperforms the benchmark models on all relevant metrics. Furthermore, Table 1 shows that the proposed method reduces the mean-rHeight by almost half compared to previous approaches. Intuitively, if we were to sequentially clear a region according to the generated risk score ranking, this metric tells us the average number of negative cells we would need to visit before the region is completely cleared. This measures how efficiently we could demine a geographic region of interest: RELand reduces the false alarms and the time required for landmine clearance by half.

Model ROC (↑) PR (↑) mean-Height (↓) mean-rHeight (↓)
LR-single (current) 86.35 (11.54) 17.07 (10.76) 3.06 (3.19) 226.79 (211.23)
LR-geo (2019, 2016) 67.62 (18.58) 5.37 (8.00) 8.09 (6.93) 573.36 (440.71)
SVM-geo (2019) 48.61 (18.09) 1.73 (1.82) 15.26 (15.66) 821.26 (729.12)
RELand (ours) 92.90 (4.43) 29.03 (22.11) 2.17 (2.48) 132.03 (133.50)
Table1 . Validation results in Colombia. Each entry is the mean (std) performance on validation folds following the block cross-validation rule. RELand is our interpretable IRM model. Full experimental results and ablation studies are available in our paper.

Hazard Cluster Identification as a Quadratic Knapsack Problem

Building a reliable prediction model to estimate landmine contamination risk is a crucial first step in data-driven prioritization of land clearance operations. However, integrating the risk maps generated by machine learning models into demining workflows requires considering the additional geographical and budgetary constraints that mine action organizations face in their ground operations. For instance, demining organizations often operate under limited budgets, allowing them to clear only a fraction of the total area under study while also covering the costs associated with mobilizing equipment and teams across the region (e.g., metal detectors, sniffing dogs, and human deminers). Moreover, if multiple regions are to be demined, there must be a secure path connecting these regions to ensure the safe movement of such demining teams. Humanitarian demining organizations need to maximize the land released back to local communities while navigating these challenges.

We propose to find which cells to prioritize for mine clearance by using a Quadratic Knapsack Problem (QKP), whose optimal solution naturally results in the identification of cohesive hazard clusters due to rewarding the program for prioritizing nearby grid cells. Formally, we use the risk scores (r_i) estimated by our trained deep learning model to compute proxies for the benefit of demining candidate grid cell (i) with centroid ((x_i,y_i)). Then, define the reward matrix (U) that captures the (additional) benefit of prioritizing both grid cells (i) and (j) as

$$u_{ij} = sqrt{r_i r_j}expleft(-lambda ||s_i – s_j||_{h}right),$$

where (||cdot||_{h}) is the standard Haversine distance, and (lambda) controls for the exponential decay of the spatial distance between two locations (s_i = (x_i, y_i)) and (s_j = (x_j, y_j)). For example, selecting a grid cell (i) for mine clearance results in a direct benefit of (u_{ii} = r_i). Note that, in our formulation, riskier cells yield greater rewards. This results in the following binary QKP with variables (z_i in {0,1}), for (iin [n]), which indicate if a grid cell (i) is selected for demining. Then, the total reward is given by (z^{T}Uz), which is maximized subject to a given budget (C in mathbb{R}_{+}) and demining costs (w_i):

$$ max_{z in mathbb{R}^n} ~ z^{T}Uz $$

$$s.t. quad sum_{i=1}^n w_i z_i leq C, quad z_i in {0, 1} quad forall i in [n].$$

Our approach rewards for geographic cohesion, ultimately finding more useful hazard clusters than a greedy solution that prioritizes the (C) grid cells with the largest estimated risk scores (Fig. 4). Moreover, our approach also incorporates realistic budget constraints, unlike standard spatial statistical approaches for geographic clustering such as Moran Local I and LISA.

Figure 4. Hazardous areas identified by RELand in our field test in Colombia. (a) Estimated risk scores from our trained DL model , (b) greedy risk clusters subject to budget constraints, and (c) QKP cohesive risk clusters with geographic pairwise interactions. Three landmines (panel (c), in white) have been found so far in one of the prioritized areas.

Tangible Impact of RELand

We are currently conducting a field study in Colombia, in partnership with the United Nations Mine Action Service and the Colombian Campaign to Ban Landmines, in two municipalities recently selected for humanitarian demining that have not been previously surveyed. We applied RELand to these regions to (i) build the enhanced dataset with rich geographic features, (ii) generate landmine contamination risk estimates by using the trained DL model, and (iii) use the predicted risk scores to identify priority hazard clusters with the QKP formulation. We worked together with the field teams of our partner NGO in Colombia to validate the hazard clusters identified by the system and to create an initial demining plan in the assigned regions. Crucially, the proposed methodology (Fig. 4c) identifies useful cohesive hazard clusters under realistic budgetary constraints. These hazard regions are more useful for demining prioritization than the sparse raw risk scores (Fig. 4a) and the greedy risk clusters (Fig. 4b), which lead to excessive mobilization of demining teams and equipment. Overall, the risk maps generated are in line with what is expected by human experts in humanitarian demining in Colombia. To date, three landmines have been found in one priority area, saving human lives. Moreover, in collaboration with UNOPS and MAPA, we have tailored and deployed the system in Afghanistan, identifying 81 hazardous areas for prioritized demining interventions, positively impacting over 4 million people across the country.

We expect to have the full results of our demining field tests within 6 months to provide a real-world validation of RELand’s capabilities in ground operations. Based on the initial positive feedback, we believe the system can support critical parts of the initial planning of humanitarian mine action, making demining operations more efficient and safer. We are actively working with UNMAS, UNOPS, and local NGOs to refine the system in its three components and prepare it for deployment in war-torn territories globally.

Aknowledgments

RELand was developed in collaboration with Cindy Zeng (UIUC), Anna Wang (CMU), Didier Alvarado (UNMAS Colombia), Francisco Moreno (CCBL), Hoda Heidari (CMU), and Fei Fang (CMU). Special thanks to UNOPS and MAPA for their partnership in our Afghanistan field tests. All errors remain mine.

References

  • Dulce Rubio, M., Zeng, S., Wang, Q., Alvarado, D., Moreno Rivera, F., Heidari, H., & Fang, F. (2024). RELand: Risk Estimation of Landmines via Interpretable Invariant Risk Minimization. ACM Journal on Computing and Sustainable Societies, 2(2), pp. 1-29. https://doi.org/10.1145/3648437.
  • Dulce Rubio, M. (2024). Identification of Hazard Clusters for Priority Landmine Clearance as a Quadratic Knapsack Problem. Doing Good with Good OR Competition, INFORMS Annual Meeting.
  • Collins, R., Fragniere, L., & Dulce Rubio, M. (2024). Advancements In Mine Action: Enhancing Remote Reporting And Analysis Through Innovative Technologies. The Journal of Conventional Weapons Destruction28(3), 7.

Read More

Build a multi-tenant generative AI environment for your enterprise on AWS

Build a multi-tenant generative AI environment for your enterprise on AWS

While organizations continue to discover the powerful applications of generative AI, adoption is often slowed down by team silos and bespoke workflows. To move faster, enterprises need robust operating models and a holistic approach that simplifies the generative AI lifecycle. In the first part of the series, we showed how AI administrators can build a generative AI software as a service (SaaS) gateway to provide access to foundation models (FMs) on Amazon Bedrock to different lines of business (LOBs). In this second part, we expand the solution and show to further accelerate innovation by centralizing common Generative AI components. We also dive deeper into access patterns, governance, responsible AI, observability, and common solution designs like Retrieval Augmented Generation.

Our solution uses Amazon Bedrock, a fully managed service that offers a choice of high-performing foundation models (FMs) from leading AI companies such as AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and Amazon through a single API via a single API, along with a broad set of capabilities to build generative AI applications with security, privacy, and responsible AI. It also uses a number of other AWS services such as Amazon API Gateway, AWS Lambda, and Amazon SageMaker.

Architecting a multi-tenant generative AI environment on AWS

A multi-tenant, generative AI solution for your enterprise needs to address the unique requirements of generative AI workloads and responsible AI governance while maintaining adherence to corporate policies, tenant and data isolation, access management, and cost control. As a result, building such a solution is often a significant undertaking for IT teams.

In this post, we discuss the key design considerations and present a reference architecture that:

  • Accelerates generative AI adoption through quick experimentation, unified model access, and reusability of common generative AI components
  • Offers tenants the flexibility to choose the optimal design and technical implementation for their use case
  • Implements centralized governance, guardrails, and controls
  • Allows for tracking and auditing model usage and cost per tenant, line of business (LOB), or FM provider

Solution overview

The proposed solution consists of two parts:

  • The generative AI gateway and
  • The tenant

The following diagram illustrates an overview of the solution.

Solution architecture

Generative AI gateway

Shared components lie in this part. Shared components refer to the functionality and features shared by all tenants. Each component in the previous diagram can be implemented as a microservice and is multi-tenant in nature, meaning it stores details related to each tenant, uniquely represented by a tenant_id. Some components are categorized in groups based on the type of functionality they exhibit.

The standalone components are:

  • The HTTPS endpoint is the entry point to the gateway. Interactions with the shared services goes through this HTTPS endpoint. This is the only entry point of the solution.
  • The orchestrator is responsible for receiving the requests forwarded by the HTTPS endpoint and invoking relevant microservices, based on the task at hand. This in itself is a microservice, inspired the Orchestrator Saga pattern in microservices.
  • The generative AI playground is a UI provided to tenants where they can run their one-time experiments, chat with several FMs, and manually test capabilities such as guardrails or model evaluation for exploration purposes.

The component groups are as follows.

  • Core services is primarily targeted to the environment administrator. It contains services used to onboard, manage, and operate the environment, for example, to onboard and off-board tenants, users, and models, assign quotas to different tenants, and authentication and authorization microservices. It also contains observability components for cost tracking, budgeting, auditing, logging, etc.
  • Generative AI model components contain microservices for foundation and custom model invocation operations. These microservices abstract communication to FMs served through Amazon Bedrock, Amazon SageMaker, or a third-party model provider.
  • Generative AI components provide functionalities needed to build a generative AI application. Capabilities such as prompt caching, prompt chaining, agents, or hybrid search are part of these microservices.
  • Responsible AI components promote the safe and responsible development of AI across tenants. They include features such as guardrails, red teaming, and model evaluation.

Tenant

This part represents the tenants using the AI gateway capabilities. Each tenant has different requirements and needs and their own application stack. They can integrate their application with the generative AI gateway to embed generative AI capabilities in their application. The environment Admin has access to the generative AI gateway and interacts with the core services.

Solution walkthrough

The following sections examine each part of the solution in more depth.

HTTPS endpoint

This serves as the entry point for the generative AI gateway. Incoming requests to the gateway go through this point. There are different approaches you can follow when designing the endpoint:

  • REST API endpoint – You can set up a REST API endpoint using services such as API Gateway where you can apply all authentication, authorization, and throttling mechanisms. API Gateway is serverless and hence automatically scales with traffic.
  • WebSockets – For long-running connections, you can use WebSockets instead of a REST interface. This implementation overcomes timeout limitations in synchronous REST requests. A WebSockets implementation keeps the connection open for multiturn or long-running conversations. API Gateway also provides a WebSocket API.
  • Load balancer – Another option is to use a load balancer that exposes an HTTPS endpoint and routes the request to the orchestrator. You can use AWS services such as Application Load Balancer to implement this approach. The advantage of using Application Load Balancer is that it can seamlessly route the request to virtually any managed, serverless or self-hosted component and can also scale well.

Tenants and access patterns

Tenants, such as LOBs or teams, use the shared services to access APIs and integrate generative AI capabilities into their applications. They can also use the playground UI to assess the suitability of generative AI for their specific use case before diving into full-fledged application development.

Here you also have the data sources, processing pipelines, vector stores, and data governance mechanisms that allow tenants to securely discover, access, andthe data they need for their specific use case. At this point, you need to consider the use case and data isolation requirements. Some applications may need to access data with personal identifiable information (PII) while others may rely on noncritical data. You also need to consider the operational characteristics and noisy neighbor risks.

Take Retrieval Augmented Generation (RAG) as an example. Depending on the use case and data isolation requirements, tenants can have a pooled knowledge base or a siloed one and implement item-level isolation or resource level isolation for the data respectively. Tenants can select data from the data sources they have access to, choose the right chunking strategy for their application, use the shared generative AI FMs for converting the data into embeddings, and store the embeddings in their vector store.

To answer user questions in real time, tenants can implement caching mechanisms to reduce latency and costs for frequent queries. Additionally, they can implement custom logic to retrieve information about previous sessions, the state of the interaction, and information specific to the end user. To generate the final response, they can again access the models and re-ranking functionality available through the gateway.

The following diagram illustrates a potential implementation of a chat-based assistant application with this approach. The tenant application uses FMs available through the generative AI gateway and its own vector store to provide personalized, relevant responses to the end user.

Retrieval Augmented Generation - Example architecture

Shared services

The following section describes the shared services groups.

Model components

The goal of this component group is to expose a unified API to tenants for accessing underlying models irrespective of where these are hosted. It abstracts invocation details and accelerates application development. It consists of one or more components depending on the number of FM providers and number and types of custom models used. These components are illustrated in the following diagram.

model components

In terms of how to offer FMs to your tenants, with AWS you have several options:

  • Amazon Bedrock is a fully managed service that offers a choice of FMs from AI companies like AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and Amazon through a single API. It’s serverless so you don’t have to manage the infrastructure. You can also bring your own customized models and deploy them to Amazon Bedrock for supported architectures.
  • SageMaker JumpStart is a machine learning (ML) hub that provides a wide range of publicly available and proprietary FMs from providers such as AI21 Labs, Cohere, Hugging Face, Meta, and Stability AI, which you can deploy to SageMaker endpoints in your own AWS account.
  • SageMaker offers SageMaker endpoints for inference where you can deploy a publicly available model, such as models from HuggingFace, or your own model.
  • You can also deploy models on AWS compute using container services such as Amazon Elastic Kubernetes Service (Amazon EKS) or self-managed approaches.

With AWS PrivateLink, you can create a private connection between your virtual private cloud (VPC) and Amazon Bedrock and SageMaker endpoints.

Generative AI application components

This group contains components linked to the unique requirements of generative AI applications. They’re illustrated in the following figure.

GenAI application components

  • Prompt catalog – Crafting effective prompts is important for guiding large language models (LLMs) to generate the desired outputs. Prompt engineering is typically an iterative process, and teams experiment with different techniques and prompt structures until they reach their target outcomes. Having a centralized prompt catalog is essential for storing, versioning, tracking, and sharing prompts. It also lets you automate your evaluation process in your pre-production environments. When a new prompt is added to the catalog, it triggers the evaluation pipeline. If it leads to better performance, your existing default prompt in the application is overridden with the new one. When you use Amazon Bedrock, Amazon Bedrock Prompt Management allows you to create and save your own prompts so you can save time by applying the same prompt to different workflows. Alternatively, you can use Amazon DynamoDB, a serverless, fully managed NoSQL database, to store your prompts.
  • Prompt chaining – Generative AI developers often use prompt chaining techniques to break complex tasks into subtasks before sending them to an LLM. A centralized service that exposes APIs for common prompt-chaining architectures to your tenants can accelerate development. You can use AWS Step Functions to orchestrate the chaining workflows and Amazon EventBridge to listen to task completion events and trigger the next step. Refer to Perform AI prompt-chaining with Amazon Bedrock for more details.
  • Agent – Tenants also often employ autonomous agents to complete complex tasks. Such agents orchestrate interactions between models, data sources, APIs, and applications. The agents component allows them to create, manage, access, and share agent implementations. On AWS, you can use the fully managed Amazon Bedrock Agents or tools of your choice such as LangChain agents or LlamaIndex agents.
  • Re-ranker – In the RAG design, a search in internal company data often returns multiple candidate outputs. A re-ranker, such as a Cohere Rerank 2 model, helps identify the best candidates based on predefined criteria. If your tenants prefer to use the capabilities of managed services such as Amazon OpenSearch Service or Amazon Kendra, this component isn’t needed.
  • Hybrid search – In RAG, you may also optionally want to implement and expose different templates for performing hybrid search that help improve the quality of the retrieved documents. This logic sits in a hybrid search component. If you use managed services such as Amazon OpenSearch Service, this component is also not required.

Responsible AI components

This group contains key components for Responsible AI, as shown in the following diagram.

responsible AI components

  • Guardrails – Guardrails help you implement safeguards in addition to the FM built-in protections. They can be applied as generic defaults for users in your organization or can be specific to each use case. You can use Amazon Bedrock Guardrails to implement such safeguards based on your application requirements and responsible AI policies. With Amazon Bedrock Guardrails, you can block undesirable topics, filter harmful content, and redact or block sensitive information such as PII and custom regular expression to protect privacy. Additionally, contextual grounding checks can help detect hallucinations in model responses based on a reference source and a user query. The ApplyGuardrail API can evaluate input prompts and model responses for FMs on Amazon Bedrock, custom FMs, and third-party FMs, enabling centralized governance across your generative AI applications.
  • Red teaming – Red teaming helps reveal model limitations that can cause bad user experiences or enable malicious intentions. LLMs can be vulnerable to security and privacy attacks such as backdoor attacks, poisoning attacks, prompt injection, jailbreaking, PII leakage attacks, membership inference attacks or gradient leakage attacks. You can set up a test application and a red team with your own employees or automate it against a known set of vulnerabilities. For example, you can test the application with known jailbreaking datasets such as these You can use the results to tailor your Amazon Bedrock Guardrails to block undesirable topics, filter harmful content, and redact or block sensitive information.
  • Human in the loop – The human-in-the-loop approach is the process of collecting human inputs across the ML lifecycle to improve the accuracy and relevancy of models. Humans can perform a variety of tasks, from data generation and annotation to model review, customization, and evaluation. With SageMaker Ground Truth, you have a self-service offering and an AWS managed In the self-service offering, your data annotators, content creators, and prompt engineers (in-house, vendor-managed, or using the public crowd) can use the low-code UI to accelerate human-in-the-loop tasks. The AWS managed offering (SageMaker Ground Truth Plus) designs and customizes an end-to-end workflow and provides a skilled AWS managed team that is trained on specific tasks and meets your data quality, security, and compliance requirements. With model evaluation in Amazon Bedrock, you can set up FM evaluation jobs that use human workers to evaluate the responses from multiple models and compare them with a ground truth response. You can set up different methods including thumbs up or down, 5-point Likert scales, binary choice buttons, or ordinal ranking.
  • Model evaluation – Model evaluation allows you to compare model outputs and choose the model best suited for downstream generative AI applications. You can use automatic model evaluations, human-in-the-loop evaluations or both. Model evaluation in Amazon Bedrock allows you to set up automatic evaluation jobs and evaluation jobs that use human workers. You can choose existing datasets or provide your own custom prompt dataset. With Amazon SageMaker Clarify, you can evaluate FMs from Amazon SageMaker JumpStart. You can set up model evaluation for different tasks such as text generation, summarization, classification, and question and answering, across different dimensions including prompt stereotyping, toxicity, factual knowledge, semantic robustness, and accuracy. Finally, you can build your own evaluation pipelines and use tools such as fmeval.
  • Model monitoring – The model monitoring service allows tenants to evaluate model performance against predefined metrics. A model monitoring solution gathers request and response data, runs evaluation jobs to calculate performance metrics against preset baselines, saves the outputs, and sends an alert in case of issues.

If you use Amazon Bedrock, you can enable model invocation logging to collect input and output data and use Amazon Bedrock evaluation to run model evaluation jobs. Alternatively, you can use AWS Lambda and implement your own logic, or use open source tools such as fmeval. In SageMaker, you can enable data capture for your SageMaker real-time endpoint and use SageMaker Clarify to run the model evaluation jobs or implement your own evaluation logic. Both Amazon Bedrock and SageMaker integrate with SageMaker Ground Truth, which helps you gather ground truth data and human feedback for model responses. AWS Step Functions can help you orchestrate the end-to-end monitoring workflow.

Core services

Core services represent a collection of administrative and management components or modules. These components are designed to provide oversight, control, and governance over various aspects of the system’s operation, resource management, user and tenant administration, and model management. These are illustrated in the following diagram.

core services

Tenant management and identity

Tenant management is a crucial aspect of multi-tenant systems, where a single instance of an application or environment serves multiple tenants or customers, each with their own isolated and secure environment. The tenant management component is responsible for managing and administering these tenants within the system.

  • Tenant onboarding and provisioning – This helps with creating a repeatable onboarding process for new tenants. It involves creating tenant-specific environments, allocating resources, and configuring access controls based on the tenant’s requirements.
  • Tenant configuration and customization – Many multi-tenant systems allow tenants to customize certain aspects of the application or environment to suit their specific needs. The tenant management component may provide interfaces or tools for tenants to configure settings, branding, workflows, or other customizable features within their isolated environments.
  • Tenant monitoring and reporting – This component is directly linked to the monitor and metering component and reports on tenant-specific usage, performance, and resource consumption. It can provide insights into tenant activity, identify potential issues, and facilitate capacity planning and resource allocation for each tenant.
  • Tenant billing and subscription management – In solutions with different pricing models or subscription plans, the tenant management component can handle billing and subscription management for each tenant based on their usage, resource consumption, or contracted service levels.

In the proposed solution, you also need an authorization flow that establishes the identity of the user making the request. With AWS IAM Identity Center, you can create or connect workforce users and centrally manage their access across their AWS accounts and applications. With Amazon Cognito, you can authenticate and authorize users from the built-in user directory, from your enterprise directory, and from other consumer identity providers. AWS Identity and Access Management (IAM) provides fine-grained access control. You can use IAM to specify who can access which FMs and resources to maintain least privilege permissions.

For example, in one common scenario with Cognito that accesses resources with API Gateway and Lambda with a user pool. In the following diagram, when your user signs in to an Amazon Cognito user pool, your application receives JSON Web Tokens (JWTs). You can use groups in a user pool to control permissions with API Gateway by mapping group membership to IAM roles. You can submit your user pool tokens with a request to API Gateway for verification by an Amazon Cognito authorizer Lambda function. For more information, see Using API Gateway with Amazon Cognito user pools.

It is recommended that you don’t use API keys for authentication or authorization to control access to your APIs. Instead, use an IAM role, a Lambda authorizer, or an Amazon Cognito user pool.

Model onboarding

A key aspect of the generative AI gateway is allowing controlled access to foundation and custom models across tenants. For FMs available through Amazon Bedrock, the model onboarding component maintains an allowlist of approved models that tenants can access. You can use a service such as Amazon DynamoDB to track allowlisted models. Similarly, for custom models deployed on Amazon SageMaker, the component tracks which tenants have access to which model versions through entries in the DynamoDB registry table.

To enforce access control, you can use AWS Lambda authorizers with Amazon API Gateway. When a tenant application calls the model invocation API, the Lambda authorizer verifies the tenant’s identity and checks if they have permission to access the requested model based on the DynamoDB registry table. If access is permitted, temporary credentials are issued, which scope down the tenant’s permissions to just the allowed model(s). This prevents tenants from accessing models they shouldn’t have access to. The authorizer logic can be customized based on an organization’s model access policies and governance requirements.

This approach supports model end of life. By managing the model from the allowlist in the DynamoDB registry table for all or selected tenants, models not included aren’t usable automatically, with no further code changes required in the solution.

Model registry

A model registry helps manage and track different versions of custom models. Services such as Amazon SageMaker Model Registry and Amazon DynamoDB help track available models, associated generated model artifacts, and lineage. A model registry offers the following:

  1. Version control – To track different versions of the generative AI models.
  2. Model lineage and provenance – To track the lineage and provenance of each model version, including information about the training data, hyperparameters, model architecture, and other relevant metadata that describes the model’s origin and characteristics.
  3. Model deployment and rollback – To facilitate the deployment and usage of new model versions into production environments and the rollback to previous versions if necessary. This makes sure that models can be updated or reverted seamlessly without disrupting the system’s operation.
  4. Model governance and compliance – To verify that model versions are properly documented, audited, and conform to relevant policies or regulations. This is particularly useful in regulated industries or environments with strict compliance requirements.

Observability

Observability is crucial for monitoring the health of your application, troubleshooting issues, usage of FMs, and optimizing performance and costs.

observability components

Logging and monitoring

Amazon CloudWatch is a powerful monitoring and observability service that allows you to collect and analyze logs from your application components, including API Gateway, Amazon Bedrock, Amazon SageMaker, and custom services. Using CloudWatch to capture tenant identity in the logs across the whole stack helps you gain insights into the performance and health of your generative AI gateway down to the tenant level and proactively identify and resolve issues before they escalate. You can also set up alarms to get notified in case of unexpected behavior. Both Amazon SageMaker and Amazon Bedrock are integrated with AWS CloudTrail.

Metering

Metering helps collect, aggregate, and analyze operational and usage data and performance metrics from different parts of the solution. In systems that offer pay-per-use or subscription-based models, metering is crucial for accurately measuring and reporting resource consumption for billing purposes across the different tenants.

In this solution, you need to track the usage of FMs to effectively manage costs and optimize resource utilization. Collecting information related to the models used, number of tokens provided as input, tokens generated as output, AWS Region used, and applying tags related to the team helps you streamline the cost allocation and billing processes. You can log structured data during interactions with the FMs and collect this usage information. The following diagram shows an implementation where the Lambda function logs per tenant information in Amazon CloudWatch and invokes Amazon Bedrock. The invocation generates an AWS CloudTrail event.

metering components

Auditing

You can use an AWS Lambda function to aggregate the data from Amazon CloudWatch and store it in S3 buckets for long-term storage and further analysis. Amazon S3 provides a highly durable, scalable, and cost-effective object storage solution, making it an ideal choice for storing large volumes of data. For implementation details, refer to part 1 of this series, Build an internal SaaS service with cost and usage tracking for foundation models on Amazon Bedrock.

auditing components

Once the data is in Amazon S3, you can use AWS analytics services such as Amazon Athena, AWS Glue Data Catalog, and Amazon QuickSight to uncover patterns in the cost and usage data, generate reports, visualize trends, and make informed decisions about resource allocation, budget forecasting, and cost optimization strategies. With AWS Glue Data Catalog, a centralized metadata repository, and Amazon Athena, an interactive query service, you can run one-time SQL queries directly on the data stored in Amazon S3. The following example describes usage and cost per model per tenant in Athena.

using Amazon Athena for cost tracking

Scaling across the enterprise

The following are some design considerations for when you scale this solution across hundreds of LOBs and teams within an organization.

  • Account limits – So far, we have discussed how to deploy the gateway solution in a single AWS account. As teams rapidly onboard to the gateway and expand their usage of LLMs, this might result in various components hitting their AWS account limits and can quickly become a bottleneck. We recommend deploying the generative AI gateway to more than one AWS accounts where each AWS account corresponds to one LOB. The reasoning behind this suggestion is, generally, the LOBs in large enterprises are quite autonomous and can each have tens to hundreds of teams. In addition, they may have strict data privacy policies which restricts them from sharing the data with other LOBs. In addition to this account, each LOB may have their non-prod AWS account as well where this gateway solution is deployed for testing and integration purposes.
  • Production and non-production workloads – In most cases, tenant teams will want to use this gateway across their development, test, and production environments. Although it largely depends on an organization’s operating model, our recommendation is to have a dedicated development, test, and production environment for the gateway as well, so the teams can experiment freely without overloading the production gateway or polluting it with non-production data. This offers the additional benefit that you can set the limits for non-production gateways lower than those in production.
  • Handling RAG data components – For implementing RAG solutions, we suggest keeping all the data-related components on the tenant’s end. Every tenant will have their own data constraints, update cycle, format, terminologies, and permission groups. Assigning the responsibility of managing data sources to the gateway may hinder scalability because the gateway can’t accommodate the unique requirements of each tenant’s data sources and most likely will end up serving the lowest common denominator. Hence, we recommend having the data sources and related components managed on the tenant’s side.
  • Avoid reinventing the wheel – With this solution, you can build and manage your own components for model evaluation, guardrails, prompt catalogue, monitoring, and more. Services such as Amazon Bedrock provide the capabilities you need to build generative AI applications with security, privacy, and responsible AI right from the start. Our recommendation is to take a balanced approach and, wherever possible, use AWS native capabilities to reduce operational costs.
  • Keeping the generative AI gateway thin – Our suggestion is to keep this gateway thin in terms of storing business logic. The gateway shouldn’t add any business rules for any specific tenant and should avoid storing any kind of tenant specific data apart from operational data already discussed in the post.

Conclusion

A generative AI multi-tenant architecture helps you maintain security, governance, and cost controls while scaling the use of generative AI across multiple use cases and teams. In this post, we presented a reference multi-tenant architecture to help you accelerate generative AI adoption. We showed how to standardize common generative AI components and how to expose them as shared services. The proposed architecture also addressed key aspects of governance, security, observability, and responsible AI. Finally, we discussed key considerations when scaling this architecture to hundreds of teams.

If you want to read more about this topic, check out also the following resources:

Let us know what you think in the comments section!


About the authors

Anastasia Tzeveleka is a Senior Generative AI/ML Specialist Solutions Architect at AWS. As part of her work, she helps customers across EMEA build foundation models and create scalable generative AI and machine learning solutions using AWS services.

Hasan Poonawala is a Senior AI/ML Specialist Solutions Architect at AWS, working with Healthcare and Life Sciences customers. Hasan helps design, deploy and scale Generative AI and Machine learning applications on AWS. He has over 15 years of combined work experience in machine learning, software development and data science on the cloud. In his spare time, Hasan loves to explore nature and spend time with friends and family.

Bruno Pistone is a Senior Generative AI and ML Specialist Solutions Architect for AWS based in Milan. He works with large customers helping them to deeply understand their technical needs and design AI and Machine Learning solutions that make the best use of the AWS Cloud and the Amazon Machine Learning stack. His expertise include: Machine Learning end to end, Machine Learning Industrialization, and Generative AI. He enjoys spending time with his friends and exploring new places, as well as travelling to new destinations

Vikesh Pandey is a Principal Generative AI/ML Solutions architect, specialising in financial services where he helps financial customers build and scale Generative AI/ML platforms and solution which scales to hundreds to even thousands of users. In his spare time, Vikesh likes to write on various blog forums and build legos with his kid.

Antonio Rodriguez is a Principal Generative AI Specialist Solutions Architect at Amazon Web Services. He helps companies of all sizes solve their challenges, embrace innovation, and create new business opportunities with Amazon Bedrock. Apart from work, he loves to spend time with his family and play sports with his friends.

Read More

Enhance customer support with Amazon Bedrock Agents by integrating enterprise data APIs

Enhance customer support with Amazon Bedrock Agents by integrating enterprise data APIs

Generative AI has transformed customer support, offering businesses the ability to respond faster, more accurately, and with greater personalization. AI agents, powered by large language models (LLMs), can analyze complex customer inquiries, access multiple data sources, and deliver relevant, detailed responses.

In this post, we guide you through integrating Amazon Bedrock Agents with enterprise data APIs to create more personalized and effective customer support experiences. Although the principles discussed are applicable across various industries, we use an automotive parts retailer as our primary example throughout this post.

By the end of this post, you’ll have a clear understanding of how to do the following:

  • Use Amazon Bedrock Agents to create intelligent, context-aware customer support bots
  • Integrate enterprise data sources, such as inventory management and catalog systems, with agents using AWS Lambda
  • Build customized chat interfaces using the Amazon Bedrock Agents API
  • Implement a solution that can instantly cross-reference product specifications with catalogs, check real-time inventory, and provide detailed information to the end-user

Solution overview

To illustrate the potential of this technology, consider an automotive parts retailer. In this industry, finding the right components can be challenging, because it often involves navigating extensive catalogs and complex compatibility requirements. An automotive retailer might use inventory management APIs to track stock levels and catalog APIs for vehicle compatibility and specifications. Access to car manuals and technical documentation helps the agent provide additional context for curated guidance, enhancing the quality of customer interactions.

The solution presented in this post takes approximately 15–30 minutes to deploy and consists of the following key components:

  • Amazon OpenSearch Service Serverless maintains three indexes: the inventory index, the compatible parts index, and the owner manuals index. These indexes enable efficient searching and retrieval of part data and vehicle information, providing quick and accurate results.
  • Amazon Bedrock Agents coordinates interactions between foundation models (FMs), knowledge bases, and user conversations. The agents also automatically call APIs to perform actions and access knowledge bases to provide additional information.
  • Amazon Bedrock Knowledge Bases enables you to use Retrieval Augmented Generation (RAG), a technique that enhances responses from LLMs by incorporating information from a data store. By setting up a knowledge base with your data sources, your application can query it to provide answers, either through direct quotes from the sources or through naturally generated responses based on the query results.
  • A web application serves as the frontend interface where users can initiate parts lookup requests.

Ingestion flow

The ingestion flow prepares and stores the necessary data for the AI agent to access. The following diagram illustrates how it works.

Workflow diagram showing Ingestion process from Amazon S3 into Bedrock Knowledge Bases

The workflow includes the following steps:

  1. Documents (owner manuals) are uploaded to an Amazon Simple Storage Service (Amazon S3) bucket.
  2. Amazon Bedrock Knowledge Bases ingests these documents:
    1. The knowledge base is configured to use the S3 bucket as a data source.
    2. The data source is synchronized and the knowledge base detects new, modified, or deleted documents in the S3 bucket and updates accordingly.
    3. The documents are chunked into smaller segments for more effective processing. This solution uses fixed-size chunking, where you can configure the desired chunk size by specifying the number of tokens per chunk and an overlap percentage.
  3. Each chunk is embedded by using an embedding model such as Cohere Embed on Amazon Bedrock to create vector representations (embeddings) of the text.
  4. The embeddings are stored in the Amazon OpenSearch Service owner manuals index. OpenSearch Service is used as the vector store for efficient similarity searching. The embeddings, along with metadata about the source documents, are indexed for quick retrieval.

User interaction flow

The following diagram illustrates the user interaction flow.

Architecture digram showing agent setup with AWS Lambda, OpenSearch and Knowledge Bases

  1. A user interacts with the Car Parts Agent through a web application interface. They can ask questions like “What wiper blades fit a 2021 Honda CR-V?” or ”Tell me about part number 76622-T0A-A01.”
  2. The web application sends the user’s query to the Amazon Bedrock agent using the InvokeAgent API. The agent, using Anthropic’s Claude 3 Sonnet, interprets the user’s query and determines the best course of action through chain-of-thought (CoT) reasoning. At this stage, the agent employs guardrails to make sure it stays within its defined scope and capabilities. Through a runtime process that includes preprocessing and postprocessing steps, the agent categorizes the user’s input. This allows it to handle out-of-scope questions or potentially harmful inputs appropriately, without attempting to answer beyond its capabilities or knowledge base. The agent then analyzes the query to extract key information such as vehicle details, part numbers, or general automotive topics. If the query is within scope, the agent proceeds; if not, it provides a response indicating it can’t assist with that particular request.
  3. For general inquiries, the agent consults its knowledge base in Amazon Bedrock, which includes information from various car manuals. This allows the agent to provide context and general information about car parts and systems.
  4. For specific part inquiries, the agent consults the action groups available to the agent and invokes the correct action (API) to retrieve relevant information. This invocation happens when the agent determines that it needs to run a specific action based on the user input.
    1. The Lambda function runs the database query against the appropriate OpenSearch Service indexes, searching for exact matches or using fuzzy matching for partial information. It can access the inventory index for specific part details or the compatible parts index for compatibility information.
    2. The Lambda function processes the OpenSearch Service results and formats them for the Amazon Bedrock agent.
  5. The Amazon Bedrock agent takes the formatted results and generates a human-readable response, combining database information with its general knowledge to provide comprehensive answers.

The following diagram illustrates the workflow of the agent.

Flow chart of user query processing cycle from input through response generation and feedback

This diagram illustrates the agent’s workflow from user query to response generation, integrating knowledge base and API data to provide comprehensive answers and handle follow-up questions.

Developer tools

The solution also uses the following developer tools:

  • AWS Powertools for Lambda – This is a suite of utilities for Lambda functions that generates OpenAPI schemas from your Lambda function code. It provides annotations for business logic, descriptions, and parameter validations, automatically producing JSON-serialized OpenAPI schemas for use with Amazon Bedrock Agents.
  • AWS Generative AI Constructs Library – This is an open source extension of the AWS Cloud Development Kit (AWS CDK) that offers multi-service, well-architected patterns for quickly defining generative AI solutions. It provides constructs to help developers build generative AI applications using pattern-based definitions for your infrastructure.

Prerequisites

You should have the following prerequisites:

Deploy the solution

The following steps outline the process to deploying the solution using the AWS CDK. The complete source code for this solution is available in the GitHub repository.

  1. Open your terminal and run the following commands to clone the GitHub repository to your local machine:
    git clone https://github.com/aws-samples/bedrock-agent-carpart-lookup.git
    cd bedrock-agent-carpart-lookup
  2. Create and activate a Python virtual environment:
    python -m venv .venv
    source .venv/bin/activate  # On Windows, use .venvScriptsactivate
    
  3. Install the required Python packages:
    pip install -r requirements.txt
    
  4. Use the AWS CDK CLI to deploy the solution:
    cdk deploy

During deployment, you may be prompted to approve IAM role creations and security changes. Review and approve these if you’re comfortable with the permissions. After deployment, the AWS CDK CLI will output the web application URL. Make note of this URL (as shown in following screenshot) to access and test the agent.

After you deploy the solution, you can verify the created resources on the Amazon Bedrock console. On the Agents page, you’ll notice a new agent called car-parts-agent.

Effective agent instructions are crucial for optimizing the performance of AI-powered assistants. A well-structured set of instructions should encompass several key components:

  • Agent role – Define the assistant’s purpose, such as serving as a Car Parts Assistant that helps users find compatible parts and automotive information
  • Agent actions – Outline primary tasks, such as identifying parts based on vehicle details, verifying compatibility, and providing technical specifications
  • Agent guidelines – Establish rules for interaction, prioritizing accuracy and safety, clearly stating uncertainties, and using actions for searches
  • Agent guardrails – Implement limits to make sure the agent operates safely and effectively, using relevant automotive knowledge to enhance user support

For example, the agent we deployed has been preconfigured with the following instruction:

You are an Car Parts Assistant, helping users find compatible parts and providing automotive information. Your main tasks are: Part Identification: Find specific parts based on vehicle details (make, model, year). Assist with partial information. Compatibility Checks: Verify if parts are compatible with given vehicles. Technical Info: Provide part specifications, features, and explain component functions. Use database functions for searches and compatibility checks. Supplement with automotive knowledge for comprehensive help. Your goal is to assist effectively while ensuring users make informed decisions about their vehicle parts. Always prioritize accuracy and safety. State uncertainties clearly.

Role Actions Guidelines Guardrails

The agent has two main components:

  • Action group – An action group named CarpartsApi is created, and the actions it can perform are defined using an OpenAPI schema. Optionally, you can use Powertools for AWS Lambda to simplify the process of generating the OpenAPI schema. For more information, refer to the PowerTools documentation on Amazon Bedrock Agents. The OpenAPI schema used by this agent can be viewed on the following GitHub repo. The action group is then associated with a Lambda function containing the business logic for these actions.
  • Knowledge base – This repository enhances the agent’s responses using RAG in Amazon Bedrock. It contains information from car manuals and technical documentation. When associating a knowledge base with an agent, you can optionally provide a description on how the agent can use the knowledge base. For this demo, we use the following description for the knowledge base:

    This knowledge base contains manuals and technical documentation about various car makes from manufacturers such as Honda, Tesla, Ford, Subaru, Kia, Toyota etc.

    Instructions

The agent employs CoT reasoning to process user queries, analyzing input against its instructions and evaluating actions based the OpenAPI provided and knowledge base description. When required information is missing, as determined by the OpenAPI schema’s specifications, the agent formulates questions to elicit necessary data from the user. This analysis and information gathering leads to a logical sequence of steps, including API calls and knowledge base queries. The resulting observation enhances the prompt for the FM, which then determines and runs the most effective actions.

For this post, we use the AWS CDK and the AWS Generative AI Constructs Library to create the Amazon Bedrock agent. This approach enables version-controlled, reproducible infrastructure as code (IaC). Alternatively, you can create agents using the AWS CLI or AWS Management Console. For more details, see create and configure agent manually.

Test and validate the solution

After you deploy the solution, you can test the agent either on the Amazon Bedrock console or through the application URL noted earlier. In the application, you can ask questions like “What wiper blades fit a 2021 Honda CR-V?” or “Tell me about part number 76622-T0A-A01.”

To illustrate the Amazon Bedrock agent’s reasoning capabilities, let’s examine the agent’s response to a user question “What wiper blades fit a 2021 Honda CR-V?” The workflow would consist of the following steps:

  1. The agent analyzes the query, identifying key elements: Honda (make), CR-V (model), 2021 (year), and wiper blades (part category).
  2. It then reasons about which API to use, selecting the get_compatible_parts function as the most appropriate for vehicle-specific part queries.
  3. The agent maps the query elements to the function’s parameters, demonstrating its understanding of how to translate natural language to API calls.
  4. The agent calls the carPartsApi/get_compatible_parts action, which queries the OpenSearch Service index for compatible parts. When an action group using a Lambda function is invoked, Amazon Bedrock sends a Lambda input event using a general format.
  5. Upon receiving a response, the agent interprets the data, identifying relevant wiper blade options. The agent formulates a natural language response, recommending specific wiper blades. Similarly, Amazon Bedrock expects a response from your Lambda function to match a pre-defined format.

We get the following natural language response:

Based on the compatible parts search for a 2021 Honda CR-V in the “Wipers” category, here are the recommended wiper blade options:

  • Wiper Blade Set (Part #76622-TLA-A11) – This is a genuine Honda wiper blade set that includes the driver and passenger side wiper blades for the CR-V.
  • Rear Wiper Blade (Part #76630-TLA-A01) – This is the rear wiper blade specifically for the 2021 CR-V.
  • You can also consider individual wiper blades like the Driver Side Wiper Blade (Part #76622-T0A-A01) or the Passenger Side Wiper Arm (Part #76630-T0A-A01) if you only need to replace one side.

This process showcases how Amazon Bedrock agents can intelligently integrate enterprise data APIs with AI-driven reasoning to provide accurate, context-aware responses to customer queries.

Clean up

To avoid future costs and clean up resources after you’re done exploring the solution, delete the resources you created by running the following command from your terminal (from the project directory):

cdk destroy

Key considerations

When implementing Amazon Bedrock Agents, consider the following factors to facilitate optimal performance and scalability:

  • Agent design – Follow these recommendations when designing your agent:
    • Keep instructions focused and clear, with specific responsibilities for the agent
    • For complex use cases, consider multiple specialized agents rather than overloading a single one
    • Explore different FMs to find the best fit for your needs, considering both behavior and cost
  • Action management – Consider the following recommendations for action management:
    • Define actions carefully, including only those that the agent should reliably perform
    • Use clear, descriptive names for actions to help the agent determine their relevance
    • Avoid overlapping actions to prevent confusion and conflicts during operation
  • Testing – Make sure your testing includes the following steps:
    • Establish clear testing protocols
    • Identify common use case inputs and set accuracy targets
    • Define edge case inputs and agree on acceptable accuracy levels
    • Determine out-of-domain inputs where the agent should not respond
    • Automate tests and run them with system changes to verify consistency and reliability
  • Performance optimization – Consider the following performance optimizations:
    • Break down complex operations into smaller actions to enhance response time and error handling
    • Implement a “fail fast” principle for invalid queries, allowing more time for complex tasks
  • Security and compliance – Use Amazon Bedrock Guardrails to prevent the agent from generating harmful content or making unauthorized actions
  • Cost management – Monitor usage-based pricing for token processing and storage, facilitating efficient resource allocation and cost management

Conclusion

Integrating enterprise data APIs with Amazon Bedrock Agents offers a powerful solution for streamlining customer support, as demonstrated in the automotive parts industry. This AI-driven approach enables rapid, accurate responses to complex queries, seamlessly integrates multiple data sources, and reduces staff workload while enhancing customer experience through context-aware interactions.

The solution discussed in this post can elevate customer support across various industries. By using Amazon Bedrock agents, organizations can create more efficient, accurate, and satisfying support experiences tailored to their specific needs. To explore how AI agents can transform your own support operations, refer to Automate tasks in your application using conversational agents.


About the Authors

Deepak Kovvuri is a Senior Solutions Architect supporting Automotive and Manufacturing Customers at AWS in the US Northeast. He has over 6 years of experience in helping customers architecting a DevOps strategy for their cloud workloads. Deepak specializes in CI/CD, Systems Administration, Infrastructure as Code and Container Services. He holds an Masters in Computer Engineering from University of Illinois at Chicago.

Kingston Bosco is a Senior Solutions Architect for Global Strategic Partners at AWS. He designs and implements solutions that optimize DevOps workflows, automate cloud operations, and improve infrastructure management for customers. He holds a Master’s in Information Systems. In his free time, he enjoys hiking with his dogs and playing soccer.

Read More

Welcome to GeForce NOW Performance: Priority Members Get Instant Upgrade

Welcome to GeForce NOW Performance: Priority Members Get Instant Upgrade

This GFN Thursday, the GeForce NOW Priority membership is getting enhancements and a fresh name to go along with it. The new Performance membership offers more GeForce-powered premium gaming — at no change in the monthly membership cost.

Gamers having a hard time deciding between the Performance and Ultimate memberships can take them both for a spin with a Day Pass, now 25% off for a limited time. Day Passes give access to 24 continuous hours of powerful cloud gaming.

In addition, seven new games are available this week, joining the over 2,000 games in the GeForce NOW library.

Time for a Glow Up

The Performance membership keeps all the same great gaming benefits and now provides members with an enhanced streaming experience at no additional cost.

Performance membership on GeForce NOW
Say hello to the Performance membership.

Performance members can stream at up to 1440p — an increase from the previous 1080p resolution — and experience games in immersive, ultrawide resolutions. They can also save their in-game graphics settings across streaming sessions, including for NVIDIA RTX features in supported titles.

All current Priority members are automatically upgraded to Performance and can take advantage of the upgraded streaming experience today.

Performance members will connect to GeForce RTX-powered gaming rigs for up to 1440p resolution. Ultimate members continue to receive the top streaming experience: connecting to GeForce RTX 4080-powered gaming rigs with up to 4K resolution and 120 frames per second, or 1080p and 240 fps in Competitive mode for games with support for NVIDIA Reflex technology.

Gamers playing on the free tier will now see they’re streaming from basic rigs, with varying specs that offer entry-level cloud gaming and are optimized for capacity.

Account portal on GeForce NOW
Time to play.

At the start of next year, GeForce NOW will roll out a 100-hour monthly playtime allowance to continue providing exceptional quality and speed — as well as shorter queue times — for Performance and Ultimate members. This ample limit comfortably accommodates 94% of members, who typically enjoy the service well within this timeframe. Members can check out how much time they’ve spent in the cloud through their account portal (see screenshot example above).

Up to 15 hours of unused playtime will automatically roll over to the next month for members, and additional hours can be purchased at $2.99 for 15 additional hours of Performance, or $5.99 for 15 additional Ultimate hours.

Loyal Member Benefit

To thank the GFN community for joining the cloud gaming revolution, GeForce NOW is offering active paid members as of Dec. 31, 2024, the ability to continue with unlimited playtime for a full year until January 2026.

New members can lock in this feature by signing up for GeForce NOW before Dec. 31, 2024. As long as a member’s account remains uninterrupted and in good standing, they’ll continue to receive unlimited playtime for all of 2025.

Don’t Pass This Up

For those looking to try out the new premium benefits and all Performance and Ultimate memberships have to offer, Day Passes are 25% off for a limited time.

Whether with the newly named Performance Day Pass at $2.99 or the Ultimate Day Pass at $5.99, members can unlock 24 hours of uninterrupted access to powerful NVIDIA GeForce RTX-powered cloud gaming servers.

Another new GeForce NOW feature lets users apply the value of their most recently purchased Day Pass toward any monthly membership if they sign up within 48 hours of the completion of their Day Pass.

Day Pass Sale on GeForce NOW
Quarter the price, full day of fun.

Dive into a vast library of over 2,000 games with enhanced graphics, including NVIDIA RTX features like ray tracing and DLSS. With the Ultimate Day Pass, snag a taste of GeForce NOW’s highest-performing membership tier and enjoy up to 4K resolution 120 fps or 1080p 240 fps across nearly any device. It’s an ideal way to experience elevated GeForce gaming in the cloud.

Thrilling New Games

Members can look for the following games available to stream in the cloud this week:

  • Planet Coaster 2 (New release on Steam, Nov. 6)
  • Teenage Mutant Ninja Turtles: Splintered Fate (New release on Steam, Nov. 6)
  • Empire of the Ants (New release on Steam, Nov. 7)
  • Unrailed 2: Back on Track (New release on Steam, Nov. 7)
  • TCG Card Shop Simulator (Steam)
  • StarCraft II (Xbox, available on PC Game Pass, Nov. 5. Members need to enable access.)
  • StarCraft Remastered (Xbox, available on PC Game Pass, Nov. 5. Members need to enable access.)

What are you planning to play this weekend? Let us know on X or in the comments below.

Read More

Unleash the power of generative AI with Amazon Q Business: How CCoEs can scale cloud governance best practices and drive innovation

Unleash the power of generative AI with Amazon Q Business: How CCoEs can scale cloud governance best practices and drive innovation

This post is co-written with Steven Craig from Hearst. 

To maintain their competitive edge, organizations are constantly seeking ways to accelerate cloud adoption, streamline processes, and drive innovation. However, Cloud Center of Excellence (CCoE) teams often can be perceived as bottlenecks to organizational transformation due to limited resources and overwhelming demand for their support.

In this post, we share how Hearst, one of the nation’s largest global, diversified information, services, and media companies, overcame these challenges by creating a self-service generative AI conversational assistant for business units seeking guidance from their CCoE. With Amazon Q Business, Hearst’s CCoE team built a solution to scale cloud best practices by providing employees across multiple business units self-service access to a centralized collection of documents and information. This freed up the CCoE to focus their time on high-value tasks by reducing repetitive requests from each business unit.

Readers will learn the key design decisions, benefits achieved, and lessons learned from Hearst’s innovative CCoE team. This solution can serve as a valuable reference for other organizations looking to scale their cloud governance and enable their CCoE teams to drive greater impact.

The challenge: Enabling self-service cloud governance at scale

Hearst undertook a comprehensive governance transformation for their Amazon Web Services (AWS) infrastructure. The CCoE implemented AWS Organizations across a substantial number of business units. These business units then used AWS best practice guidance from the CCoE by deploying landing zones with AWS Control Tower, managing resource configuration with AWS Config, and reporting the efficacy of controls with AWS Audit Manager. As individual business units sought guidance on adhering to the AWS recommended best practices, the CCoE created written directives and enablement materials to facilitate the scaled adoption across Hearst.

The existing CCoE model had several obstacles slowing adoption by business units:

  • Extreme demand – The CCoE team was becoming a bottleneck, unable to keep up with the growing demand for their expertise and guidance. The team was stretched thin, and the traditional approach of relying on human experts to address every question was impeding the pace of cloud adoption for the organization.
  • Limited scalability – As the volume of requests increased, the CCoE team couldn’t disseminate updated directives quickly enough. Manually reviewing each request across multiple business units wasn’t sustainable.
  • Inconsistent governance – Without a standardized, self-service mechanism to access the CCoE teams’ expertise and disseminate guidance on new policies, compliance practices, or governance controls, it was difficult to maintain consistency based on the CCoE best practices across each business unit.

To address these challenges, Hearst’s CCoE team recognized the need to quickly create a scalable, self-service application that could empower the business units with more access to updated CCoE best practices and patterns to follow.

Overview of solution

To enable self-service cloud governance at scale, Hearst’s CCoE team decided to use the power of generative AI with Amazon Q Business to build a conversational assistant. The following diagram shows the solution architecture:

Hearst Arch Diagram

The key steps Hearst took to implement Amazon Q Business were:

  1. Application deployment and authentication – First, the CCoE team deployed Amazon Q Business and integrated AWS IAM Identity Center with their existing identity provider (using Okta in this case) to seamlessly manage user access and permissions between their existing identity provider and Amazon Q Business.
  2. Data source curation and authorization – The CCoE team created several Amazon Simple Storage Service (Amazon S3) buckets to store their curated content, including cloud governance best practices, patterns, and guidance. They set up a general bucket for all users and specific buckets tailored to each business unit’s needs. User authorization for documents within the individual S3 buckets were controlled through access control lists (ACLs). You add access control information to a document in an Amazon S3 data source using a metadata file associated with the document. This made sure end users would only receive responses from documents they were authorized to view. With the Amazon Q Business S3 connector, the CCoE team was able to sync and index their data in just a few clicks.
  3. User access management – With the data source and access controls in place, the CCoE team then set up user access on a business unit by business unit basis, considering various security, compliance, and custom requirements. As a result, the CCoE could deliver a personalized experience to each business unit.
  4. User interface development – To provide a user-friendly experience, Hearst built a custom web interface so employees could interact with the Amazon Q Business assistant through a familiar and intuitive interface. This encouraged widespread adoption and self-service among the business units.
  5. Rollout and continuous improvement – Finally, the CCoE team shared the web experience with the various business units, empowering employees to access the guidance and best practices they needed through natural language interactions. Going forward, the team enriched the knowledge base (S3 buckets) and implemented a feedback loop to facilitate continuous improvement of the solution.

For Hearst’s CCoE team, Amazon Q Business was the quickest way to use generative AI on AWS, with minimal risk and less upfront technical complexity.

  • Speed to value was an important advantage because it allowed the CCoE to get these powerful generative AI capabilities into the hands of employees as quickly as possible, unlocking new levels of scalability, efficiency, and innovation for cloud governance consistency across the organization.
  • This strategic decision to use a managed service at the application layer, such as Amazon Q Business, enabled the CCoE to deliver tangible value for the business units in a matter of weeks. By opting for the expedited path to using generative AI on AWS, Hearst was never bogged down in the technical complexities of developing and managing their own generative AI application.

The results: Decreased support requests and increased cloud governance consistency

By using Amazon Q Business, Hearst’s CCoE team achieved remarkable results in empowering self-service cloud governance across the organization. The initial impact was immediate—within the first month, the CCoE team saw a 70% reduction in the volume of requests for guidance and support from the various business units. This freed up the team to focus on higher-value initiatives instead of getting bogged down in repetitive, routine requests. The following month, the number of requests for CCoE support dropped by 76%, demonstrating the power of a self-service assistant with Amazon Q Business. The benefits went beyond just reduced request volume. The CCoE team also saw a significant improvement in the consistency and quality of cloud governance practices across Hearst, enhancing the organization’s overall cloud security, compliance posture, and cloud adoption.

Conclusion

Cloud governance is a critical set of rules, processes, and reports that guide organizations to follow best practices across their IT estate. For Hearst, the CCoE team sets the tone and cloud governance standards that each business unit follows. The implementation of Amazon Q Business allowed Hearst’s CCoE team to scale the governance and security that support business units depend on through a generative AI assistant. By disseminating best practices and guidance across the organization, the CCoE team freed up resources to focus on strategic initiatives, while employees gained access to a self-service application, reducing the burden on the central team. If your CCoE team is looking to scale its impact and enable your workforce, consider using the power of conversational AI through services like Amazon Q Business, which can position your team as a strategic enabler of cloud transformation.

Listen to Steven Craig share how Hearst leveraged Amazon Q Business to scale the Cloud Center of Excellence

Reading References:


About the Authors

Steven Craig is a Sr. Director, Cloud Center of Excellence. He oversees Cloud Economics, Cloud Enablement, and Cloud Governance for all Hearst-owned companies. Previously, as VP Product Strategy and Ops at Innova Solutions, he was instrumental in migrating applications to public cloud platforms and creating IT Operations Managed Service offerings. His leadership and technical solutions were key in achieving sequential AWS Managed Services Provider certifications. Steven has been AWS Professionally certified for over 8 years.

Oleg Chugaev is a Principal Solutions Architect and Serverless evangelist with 20+ years in IT, holding multiple AWS certifications. At AWS, he drives customers through their cloud transformation journeys by converting complex challenges into actionable roadmaps for both technical and business audiences.

Rohit Chaudhari is a Senior Customer Solutions Manager with over 15 years of diverse tech experience. His background spans customer success, product management, digital transformation coaching, engineering, and consulting. At AWS, Rohit serves as a trusted advisor for customers to work backwards from their business goals, accelerate their journey to the cloud, and implement innovative solutions.

Al Destefano is a Generative AI Specialist at AWS based in New York City. Leveraging his AI/ML domain expertise, Al develops and executes global go-to-market strategies that drive transformative results for AWS customers at scale. He specializes in helping enterprise customers harness the power of Amazon Q, a generative AI-powered assistant, to overcome complex challenges and unlock new business opportunities.

Read More

Integrate foundation models into your code with Amazon Bedrock

Integrate foundation models into your code with Amazon Bedrock

The rise of large language models (LLMs) and foundation models (FMs) has revolutionized the field of natural language processing (NLP) and artificial intelligence (AI). These powerful models, trained on vast amounts of data, can generate human-like text, answer questions, and even engage in creative writing tasks. However, training and deploying such models from scratch is a complex and resource-intensive process, often requiring specialized expertise and significant computational resources.

Enter Amazon Bedrock, a fully managed service that provides developers with seamless access to cutting-edge FMs through simple APIs. Amazon Bedrock streamlines the integration of state-of-the-art generative AI capabilities for developers, offering pre-trained models that can be customized and deployed without the need for extensive model training from scratch. Amazon maintains the flexibility for model customization while simplifying the process, making it straightforward for developers to use cutting-edge generative AI technologies in their applications. With Amazon Bedrock, you can integrate advanced NLP features, such as language understanding, text generation, and question answering, into your applications.

In this post, we explore how to integrate Amazon Bedrock FMs into your code base, enabling you to build powerful AI-driven applications with ease. We guide you through the process of setting up the environment, creating the Amazon Bedrock client, prompting and wrapping code, invoking the models, and using various models and streaming invocations. By the end of this post, you’ll have the knowledge and tools to harness the power of Amazon Bedrock FMs, accelerating your product development timelines and empowering your applications with advanced AI capabilities.

Solution overview

Amazon Bedrock provides a simple and efficient way to use powerful FMs through APIs, without the need for training custom models. For this post, we run the code in a Jupyter notebook within VS Code and use Python. The process of integrating Amazon Bedrock into your code base involves the following steps:

  1. Set up your development environment by importing the necessary dependencies and creating an Amazon Bedrock client. This client will serve as the entry point for interacting with Amazon Bedrock FMs.
  2. After the Amazon Bedrock client is set up, you can define prompts or code snippets that will be used to interact with the FMs. These prompts can include natural language instructions or code snippets that the model will process and generate output based on.
  3. With the prompts defined, you can invoke the Amazon Bedrock FM by passing the prompts to the client. Amazon Bedrock supports various models, each with its own strengths and capabilities, allowing you to choose the most suitable model for your use case.
  4. Depending on the model and the prompts provided, Amazon Bedrock will generate output, which can include natural language text, code snippets, or a combination of both. You can then process and integrate this output into your application as needed.
  5. For certain models and use cases, Amazon Bedrock supports streaming invocations, which allow you to interact with the model in real time. This can be particularly useful for conversational AI or interactive applications where you need to exchange multiple prompts and responses with the model.

Throughout this post, we provide detailed code examples and explanations for each step, helping you seamlessly integrate Amazon Bedrock FMs into your code base. By using these powerful models, you can enhance your applications with advanced NLP capabilities, accelerate your development process, and deliver innovative solutions to your users.

Prerequisites

Before you dive into the integration process, make sure you have the following prerequisites in place:

  • AWS account – You’ll need an AWS account to access and use Amazon Bedrock. If you don’t have one, you can create a new account.
  • Development environment – Set up an integrated development environment (IDE) with your preferred coding language and tools. You can interact with Amazon Bedrock using AWS SDKs available in Python, Java, Node.js, and more.
  • AWS credentialsConfigure your AWS credentials in your development environment to authenticate with AWS services. You can find instructions on how to do this in the AWS documentation for your chosen SDK. We walk through a Python example in this post.

With these prerequisites in place, you’re ready to start integrating Amazon Bedrock FMs into your code.

In your IDE, create a new file. For this example, we use a Jupyter notebook (Kernel: Python 3.12.0).

In the following sections, we demonstrate how to implement the solution in a Jupyter notebook.

Set up the environment

To begin, import the necessary dependencies for interacting with Amazon Bedrock. The following is an example of how you can do this in Python.

First step is to import boto3 and json:

import boto3, json

Next, create an instance of the Amazon Bedrock client. This client will serve as the entry point for interacting with the FMs. The following is a code example of how to create the client:

bedrock_runtime = boto3.client(
    service_name='bedrock-runtime',
    region_name='us-east-1'
)

Define prompts and code snippets

With the Amazon Bedrock client set up, define prompts and code snippets that will be used to interact with the FMs. These prompts can include natural language instructions or code snippets that the model will process and generate output based on.

In this example, we asked the model, “Hello, who are you?”.

To send the prompt to the API endpoint, you need some keyword arguments to pass in. You can get these arguments from the Amazon Bedrock console.

  1. On the Amazon Bedrock console, choose Base models in the navigation pane.
  1. Select Titan Text G1 – Express.
  1. Choose the model name (Titan Text G1 – Express) and go to the API request.
  1. Copy the API request:
{
"modelId": "amazon.titan-text-express-v1",
"contentType": "application/json",
"accept": "application/json",
"body": "{"inputText":"this is where you place your input text","textGenerationConfig":{"maxTokenCount":8192,"stopSequences":[],"temperature":0,"topP":1}}"
}
  1. Insert this code in the Jupyter notebook with the following minor modifications:
    • We post the API requests to keyword arguments (kwargs).
    • The next change is on the prompt. We will replace ”this is where you place your input text” by ”Hello, who are you?”
  2. Print the keyword arguments:
kwargs = {
 "modelId": "amazon.titan-text-express-v1",
 "contentType": "application/json",
 "accept": "application/json",
 "body": "{"inputText":"Hello, who are you?","textGenerationConfig":{"maxTokenCount":8192,"stopSequences":[],"temperature":0,"topP":1}}"
}
print(kwargs)

This should give you the following output:

{'modelId': 'amazon.titan-text-express-v1', 'contentType': 'application/json', 'accept': 'application/json', 'body': '{"inputText":"Hello, who are you?","textGenerationConfig":{"maxTokenCount":8192,"stopSequences":[],"temperature":0,"topP":1}}'}

Invoke the model

With the prompt defined, you can now invoke the Amazon Bedrock FM.

  1. Pass the prompt to the client:
response = bedrock_runtime.invoke_model(**kwargs)
response

This will invoke the Amazon Bedrock model with the provided prompt and print the generated streaming body object response.

{'ResponseMetadata': {'RequestId': '3cfe2718-b018-4a50-94e3-59e2080c75a3',
'HTTPStatusCode': 200,
'HTTPHeaders': {'date': 'Fri, 18 Oct 2024 11:30:14 GMT',
'content-type': 'application/json',
'content-length': '255',
'connection': 'keep-alive',
'x-amzn-requestid': '3cfe2718-b018-4a50-94e3-59e2080c75a3',
'x-amzn-bedrock-invocation-latency': '1980',
'x-amzn-bedrock-output-token-count': '37',
'x-amzn-bedrock-input-token-count': '6'},
'RetryAttempts': 0},
'contentType': 'application/json',
'body': <botocore.response.StreamingBody at 0x105e8e7a0>}

The preceding Amazon Bedrock runtime invoke model will work for the FM you choose to invoke.

  1. Unpack the JSON string as follows:
response_body = json.loads(response.get('body').read())
response_body

You should get a response as follows (this is the response we got from the Titan Text G1 – Express model for the prompt we supplied).

{'inputTextTokenCount': 6, 'results': [{'tokenCount': 37, 'outputText': 'nI am Amazon Titan, a large language model built by AWS. It is designed to assist you with tasks and answer any questions you may have. How may I help you?', 'completionReason': 'FINISH'}]}

Experiment with different models

Amazon Bedrock offers various FMs, each with its own strengths and capabilities. You can specify which model you want to use by passing the model_name parameter when creating the Amazon Bedrock client.

  1. Like the previous Titan Text G1 – Express example, get the API request from the Amazon Bedrock console. This time, we use Anthropic’s Claude on Amazon Bedrock.

{
"modelId": "anthropic.claude-v2",
"contentType": "application/json",
"accept": "*/*",
"body": "{"prompt":"\n\nHuman: Hello world\n\nAssistant:","max_tokens_to_sample":300,"temperature":0.5,"top_k":250,"top_p":1,"stop_sequences":["\n\nHuman:"],"anthropic_version":"bedrock-2023-05-31"}"
}

Anthropic’s Claude accepts the prompt in a different way (\n\nHuman:), so the API request on the Amazon Bedrock console provides the prompt in the way that Anthropic’s Claude can accept.

  1. Edit the API request and put it in the keyword argument:
    kwargs = {
      "modelId": "anthropic.claude-v2",
      "contentType": "application/json",
      "accept": "*/*",
      "body": "{"prompt":"\n\nHuman: we have received some text without any context.\nWe will need to label the text with a title so that others can quickly see what the text is about \n\nHere is the text between these <text></text> XML tags\n\n<text>\nToday I sent to the beach and saw a whale. I ate an ice-cream and swam in the sea\n</text>\n\nProvide title between <title></title> XML tags\n\nAssistant:","max_tokens_to_sample":300,"temperature":0.5,"top_k":250,"top_p":1,"stop_sequences":["\n\nHuman:"],"anthropic_version":"bedrock-2023-05-31"}"
    }
    print(kwargs)

You should get the following response:

{'modelId': 'anthropic.claude-v2', 'contentType': 'application/json', 'accept': '*/*', 'body': '{"prompt":"\n\nHuman: we have received some text without any context.\nWe will need to label the text with a title so that others can quickly see what the text is about \n\nHere is the text between these <text></text> XML tags\n\n<text>\nToday I sent to the beach and saw a whale. I ate an ice-cream and swam in the sea\n</text>\n\nProvide title between <title></title> XML tags\n\nAssistant:","max_tokens_to_sample":300,"temperature":0.5,"top_k":250,"top_p":1,"stop_sequences":["\n\nHuman:"],"anthropic_version":"bedrock-2023-05-31"}'}

  1. With the prompt defined, you can now invoke the Amazon Bedrock FM by passing the prompt to the client:
response = bedrock_runtime.invoke_model(**kwargs)
response

You should get the following output:

{'ResponseMetadata': {'RequestId': '72d2b1c7-cbc8-42ed-9098-2b4eb41cd14e', 'HTTPStatusCode': 200, 'HTTPHeaders': {'date': 'Thu, 17 Oct 2024 15:07:23 GMT', 'content-type': 'application/json', 'content-length': '121', 'connection': 'keep-alive', 'x-amzn-requestid': '72d2b1c7-cbc8-42ed-9098-2b4eb41cd14e', 'x-amzn-bedrock-invocation-latency': '538', 'x-amzn-bedrock-output-token-count': '15', 'x-amzn-bedrock-input-token-count': '100'}, 'RetryAttempts': 0}, 'contentType': 'application/json', 'body': <botocore.response.StreamingBody at 0x1200b5990>}

  1. Unpack the JSON string as follows:
response_body = json.loads(response.get('body').read())
response_body

This results in the following output on the title for the given text.

{'type': 'completion',
'completion': ' <title>A Day at the Beach</title>',
'stop_reason': 'stop_sequence',
'stop': 'nnHuman:'}

  1. Print the completion:
completion = response_body.get('completion')
completion

Because the response is returned in the XML tags as you defined, you can consume the response and display it to the client.

' <title>A Day at the Beach</title>'

Invoke model with streaming code

For certain models and use cases, Amazon Bedrock supports streaming invocations, which allow you to interact with the model in real time. This can be particularly useful for conversational AI or interactive applications where you need to exchange multiple prompts and responses with the model. For example, if you’re asking the FM for an article or story, you might want to stream the output of the generated content.

  1. Import the dependencies and create the Amazon Bedrock client:
import boto3, json
bedrock_runtime = boto3.client(
service_name='bedrock-runtime',
region_name='us-east-1'
)
  1. Define the prompt as follows:
prompt = "write an article about fictional planet Foobar"
  1. Edit the API request and put it in keyword argument as before:
    We use the API request of the claude-v2 model.
kwargs = {
  "modelId": "anthropic.claude-v2",
  "contentType": "application/json",
  "accept": "*/*",
  "body": "{"prompt":"\n\nHuman: " + prompt + "\nAssistant:","max_tokens_to_sample":300,"temperature":0.5,"top_k":250,"top_p":1,"stop_sequences":["\n\nHuman:"],"anthropic_version":"bedrock-2023-05-31"}"
}
  1. You can now invoke the Amazon Bedrock FM by passing the prompt to the client:
    We use invoke_model_with_response_stream instead of invoke_model.
response = bedrock_runtime.invoke_model_with_response_stream(**kwargs)

stream = response.get('body')
if stream:
    for event in stream:
        chunk = event.get('chunk')
        if chunk:
            print(json.loads(chunk.get('bytes')).get('completion'), end="")

You get a response like the following as streaming output:

Here is a draft article about the fictional planet Foobar: Exploring the Mysteries of Planet Foobar Far off in a distant solar system lies the mysterious planet Foobar. This strange world has confounded scientists and explorers for centuries with its bizarre environments and alien lifeforms. Foobar is slightly larger than Earth and orbits a small, dim red star. From space, the planet appears rusty orange due to its sandy deserts and red rock formations. While the planet looks barren and dry at first glance, it actually contains a diverse array of ecosystems. The poles of Foobar are covered in icy tundra, home to resilient lichen-like plants and furry, six-legged mammals. Moving towards the equator, the tundra slowly gives way to rocky badlands dotted with scrubby vegetation. This arid zone contains ancient dried up riverbeds that point to a once lush environment. The heart of Foobar is dominated by expansive deserts of fine, deep red sand. These deserts experience scorching heat during the day but drop to freezing temperatures at night. Hardy cactus-like plants manage to thrive in this harsh landscape alongside tough reptilian creatures. Oases rich with palm-like trees can occasionally be found tucked away in hidden canyons. Scattered throughout Foobar are pockets of tropical jungles thriving along rivers and wetlands.

Conclusion

In this post, we showed how to integrate Amazon Bedrock FMs into your code base. With Amazon Bedrock, you can use state-of-the-art generative AI capabilities without the need for training custom models, accelerating your development process and enabling you to build powerful applications with advanced NLP features.

Whether you’re building a conversational AI assistant, a code generation tool, or another application that requires NLP capabilities, Amazon Bedrock provides a simple and efficient solution. By using the power of FMs through Amazon Bedrock APIs, you can focus on building innovative solutions and delivering value to your users, without worrying about the underlying complexities of language models.

As you continue to explore and integrate Amazon Bedrock into your projects, remember to stay up to date with the latest updates and features offered by the service. Additionally, consider exploring other AWS services and tools that can complement and enhance your AI-driven applications, such as Amazon SageMaker for machine learning model training and deployment, or Amazon Lex for building conversational interfaces.

To further explore the capabilities of Amazon Bedrock, refer to the following resources:

Share and learn with our generative AI community at community.aws.

Happy coding and building with Amazon Bedrock!


About the Authors

Rajakumar Sampathkumar is a Principal Technical Account Manager at AWS, providing customer guidance on business-technology alignment and supporting the reinvention of their cloud operation models and processes. He is passionate about cloud and machine learning. Raj is also a machine learning specialist and works with AWS customers to design, deploy, and manage their AWS workloads and architectures.

YaduKishore Tatavarthi is a Senior Partner Solutions Architect at Amazon Web Services, supporting customers and partners worldwide. For the past 20 years, he has been helping customers build enterprise data strategies, advising them on Generative AI, cloud implementations, migrations, reference architecture creation, data modeling best practices, and data lake/warehouse architectures.

Read More

Build and deploy a UI for your generative AI applications with AWS and Python

Build and deploy a UI for your generative AI applications with AWS and Python

The emergence of generative AI has ushered in a new era of possibilities, enabling the creation of human-like text, images, code, and more. However, as exciting as these advancements are, data scientists often face challenges when it comes to developing UIs and to prototyping and interacting with their business users. Traditionally, building frontend and backend applications has required knowledge of web development frameworks and infrastructure management, which can be daunting for those with expertise primarily in data science and machine learning.

AWS provides a powerful set of tools and services that simplify the process of building and deploying generative AI applications, even for those with limited experience in frontend and backend development. In this post, we explore a practical solution that uses Streamlit, a Python library for building interactive data applications, and AWS services like Amazon Elastic Container Service (Amazon ECS), Amazon Cognito, and the AWS Cloud Development Kit (AWS CDK) to create a user-friendly generative AI application with authentication and deployment.

Solution overview

For this solution, you deploy a demo application that provides a clean and intuitive UI for interacting with a generative AI model, as illustrated in the following screenshot.

The UI consists of a text input area where users can enter their queries, and an output area to display the generated results.

The default interface is simple and straightforward, but you can extend and customize it to fit your specific needs. With Streamlit’s flexibility, you can add additional features, adjust the styling, and integrate other functionalities as required by your use case.

The solution we explore consists of two main components: a Python application for the UI and an AWS deployment architecture for hosting and serving the application securely.

The Python application uses the Streamlit library to provide a user-friendly interface for interacting with a generative AI model. Streamlit allows data scientists to create interactive web applications using Python, using their existing skills and knowledge. With Streamlit, you can quickly build and iterate on your application without the need for extensive frontend development experience.

The AWS deployment architecture makes sure the Python application is hosted and accessible from the internet to authenticated users. The solution uses the following key components:

  • Amazon ECS and AWS Fargate provide a serverless container orchestration platform for running the Python application
  • Amazon Cognito handles user authentication, making sure only authorized users can access the generative AI application
  • Application Load Balancer (ALB) and Amazon CloudFront are responsible for load balancing and content delivery, so the application is available for users worldwide
  • The AWS CDK allows you to define and provision AWS infrastructure resources using familiar programming languages like Python
  • Amazon Bedrock is a fully managed service that offers a choice of high-performing generative AI models through an API

The following diagram illustrates this architecture.

Prerequisites

As a prerequisite, you need to enable model access in Amazon Bedrock and have access to a Linux or macOS development environment. You could also use a Windows development environment, in which case you need to update the instructions in this post.

Access to Amazon Bedrock foundation models is not granted by default. Complete the following steps to enable access to Anthropic’s Claude on Amazon Bedrock, which we use as part of this post:

  1. Sign in to the AWS Management Console.
  2. Choose the us-east-1 AWS Region from the top right corner.
  3. On the Amazon Bedrock console, choose Model access in the navigation pane.
  4. Choose Manage model access.
  5. Select the model you want access to (for this post, Anthropic’s Claude). You can also select other models for future use.
  6. Choose Next and then Submit to confirm your selection.

For more information on how to manage model access, see Access Amazon Bedrock foundation models.

Set up your development environment

To get started with deploying the Streamlit application, you need access to a development environment with the following software installed:

You also need to configure the AWS CLI. One way to do it is to get your access key through the console, and use the aws configure command in your terminal to set up your credentials.

Clone the GitHub repository

Use the terminal of your development environment to enter the commands in the following steps:

  1. Clone the deploy-streamlit-app repository from the AWS Samples GitHub repository:
git clone https://github.com/aws-samples/deploy-streamlit-app.git

  1. Navigate to the cloned repository:
cd deploy-streamlit-app

Create the Python virtual environment and install the AWS CDK

Complete the following steps to set up the virtual environment and the AWS CDK:

  1. Create a new Python virtual environment (your Python version should be 3.8 or greater):
python3 -m venv .venv
  1. Activate the virtual environment:
source .venv/bin/activate
  1. Install the AWS CDK, which is in the required Python dependencies:
pip install -r requirements.txt

Configure the Streamlit application

Complete the following steps to configure the Streamlit application:

  1. In the docker_app directory, locate the config_file.py file.
  2. Open config_file.py in your editor and modify the STACK_NAME and CUSTOM_HEADER_VALUE variables:
    1. The stack name enables you to deploy multiple applications in the same account. Choose a different stack name for each application. For your first application, you can leave the default value.
    2. The custom header value is a security token that CloudFront uses to authenticate on the load balancer. You can choose it randomly, and it must be kept secret.

Deploy the AWS CDK template

Complete the following steps to deploy the AWS CDK template:

  1. From your terminal, bootstrap the AWS CDK:
cdk bootstrap
  1. Deploy the AWS CDK template, which will create the necessary AWS resources:
cdk deploy
  1. Enter y (yes) when asked if you want to deploy the changes.

The deployment process may take 5–10 minutes. When it’s complete, note the CloudFront distribution URL and Amazon Cognito user pool ID from the output.

Create an Amazon Cognito user

Complete the following steps to create an Amazon Cognito user:

  1. On the Amazon Cognito console, navigate to the user pool that you created as part of the AWS CDK deployment.
  2. On the Users tab, choose Create user.

  1. Enter a user name and password.
  2. Choose Create user.

Access the Streamlit application

Complete the following steps to access the Streamlit application:

  1. Open a new web browser window or tab and navigate to the CloudFront distribution URL from the AWS CDK deployment output.

If you have not noted this URL, you can open the AWS CloudFormation console and find it in the outputs of the stack.

  1. Log in to the Streamlit application using the Amazon Cognito user credentials you created in the previous step.

You should now be able to access and interact with the Streamlit application, which is deployed and running on AWS using the provided AWS CDK template.

This deployment is intended as a starting point and a demo. Before using this application in a production environment, you should thoroughly review and implement appropriate security measures, such as configuring HTTPS on the load balancer and following AWS best practices for securing your resources. See the README.md file in the GitHub repository for more information.

Customize the application

The aws-samples/deploy-streamlit-app GitHub repository provides a solid foundation for building and deploying generative AI applications, but it’s also highly customizable and extensible.

Let’s explore how you can customize the Streamlit application. Because the application is written in Python, you can modify it to integrate with different generative AI models, add new features, or change the UI to better align with your application’s requirements.

For example, let’s say you want to add a button to invoke the LLM answer instead of invoking it automatically when the user enters input text. Complete the following steps to modify the docker_app/app.py file:

  1. After the definition of the input_sent text input, add a Streamlit button:
# Insert this after the line starting with input_sent = …
submit_button = st.button("Get LLM Response")
  1. Change the if condition to check if the button is clicked instead of checking for input_sent:
# Replace the line `if input_sent:` by the following
if submit_button:

  1. Redeploy the application by entering the following in the terminal:
cdk deploy

The deployment should take less than 5 minutes. In the next section, we show how to test your changes locally before deploying, which will accelerate your development workflow.

  1. When the deployment is complete, refresh the webpage in your browser.

The Streamlit application will now display a button labeled Get LLM Response. When the user chooses this button, the LLM will be invoked, and the output will be displayed on the UI.

This is just one example of how you can customize the Streamlit application to meet your specific requirements. You can modify the code further to integrate with different generative AI models, add additional features, or enhance the UI as needed.

Test your changes locally before deploying

Although deploying the application using cdk deploy allows you to test your changes in the actual AWS environment, it can be time-consuming, especially during the development and testing phase. Fortunately, you can run and test your application locally before deploying it to AWS.

To test your changes locally, follow these steps:

  1. In your terminal, navigate to the docker_app directory, where the Streamlit application is located:
cd docker_app
  1. If you haven’t already, install the dependencies of the Python application. These dependencies are different from the ones of the AWS CDK application that you installed previously.
pip install -r requirements.txt
  1. Start the Streamlit server with the following command:
streamlit run app.py --server.port 8080

This will start the Streamlit application on port 8080.

You should now be able to interact with the locally running Streamlit application and test your changes without having to redeploy the application to AWS.

Remember to stop the Streamlit server (by pressing Ctrl+C in the terminal) when you’re done testing.

By testing your changes locally, you can significantly speed up the development and testing cycle, allowing you to iterate more quickly and catch issues early in the process.

Clean up

To avoid incurring additional charges, clean up the resources created during this demo:

  1. Open the terminal in your development environment.
  2. Make sure you’re in the root directory of the project and your virtual environment is activated:
cd ~/environment/deploy-streamlit-app
source .venv/bin/activate
  1. Destroy the AWS CDK stack:
cdk destroy
  1. Confirm the deletion by entering yes when prompted.

Conclusion

Building and deploying user-friendly generative AI applications no longer requires extensive knowledge of frontend and backend development frameworks. By using Streamlit and AWS services, data scientists can focus on their core expertise while still delivering secure, scalable, and accessible applications to business users.

The full code of the demo is available in the GitHub repository. It provides a valuable starting point for building and deploying generative AI applications, allowing you to quickly set up a working prototype and iterate from there. We encourage you to explore the repository and experiment with the provided solution to create your own applications.

As the adoption of generative AI continues to grow, the ability to build and deploy user-friendly applications will become increasingly important. With AWS and Python, data scientists now have the tools and resources to bridge the gap between their technical expertise and the need to showcase their models to business users through secure and accessible UIs.


About the Author

Picture of Lior PerezLior Perez is a Principal Solutions Architect on the Construction team based in Toulouse, France. He enjoys supporting customers in their digital transformation journey, using big data, machine learning, and generative AI to help solve their business challenges. He is also personally passionate about robotics and IoT, and constantly looks for new ways to use technologies for innovation.

Read More

Unearth insights from audio transcripts generated by Amazon Transcribe using Amazon Bedrock

Unearth insights from audio transcripts generated by Amazon Transcribe using Amazon Bedrock

Generative AI continues to push the boundaries of what’s possible. One area garnering significant attention is the use of generative AI to analyze audio and video transcripts, increasing our ability to extract valuable insights from content stored in audio or video files. Speech data is unique and complex, which makes it difficult to analyze and extract insights. Manually transcribing and analyzing it can be time-consuming and resource-intensive.

Existing methods for extracting insights from speech data often require tedious human transcription and review. You can use automatic voice recognition tools to convert your audio and video data to text. However, you still have to rely on manual processes for extracting specific insights and data points, or get summaries of the content. This approach is time-consuming and as organizations amass vast amounts of this content, the need for a more efficient and insightful solution becomes increasingly pressing. There is a significant opportunity to add business value given the amount of data organizations store in these formats and the valuable insights that might otherwise go undiscovered. The following are some of the new insights and capabilities that can be obtained through the use of large language models (LLM) with audio transcripts:

  • LLMs can analyze and understand the context of a conversation, not just the words spoken, but also the implied meaning, intent, and emotions. Previously, this would have required extensive human interpretation.
  • LLMs can perform advanced sentiment analysis. Previously, sentiment analysis could be captured, but LLMs can capture more emotions, such as sarcasm, ambivalence, or mixed feelings by understanding the context of the conversation.
  • LLMs can generate concise summarizations not just by extracting content, but by understanding the context of the conversation.
  • Users can now ask complex, natural language questions and receive insightful answers.
  • LLMs can infer personas or roles in a conversation, enabling targeted insights and actions.
  • LLMs can support the creation of new content based on audio assets or conversations following predetermined templates or flows.

In this post, we examine how to create business value through speech analytics with some examples focused on the following:

  • Automatically summarizing, categorizing, and analyzing marketing content such as podcasts, recorded interviews, or videos, and creating new marketing materials based on those assets
  • Automatically extracting key points, summaries, and sentiment from a recorded meeting (such as an earnings call)
  • Transcribing and analyzing contact center calls to improve customer experience.

The first step in getting these audio data insights involves transcribing the audio file using Amazon Transcribe. Amazon Transcribe is a machine learning (ML) based managed service that automatically converts speech to text, enabling developers to seamlessly integrate speech-to-text capabilities into their applications. It also recognizes multiple speakers, automatically redacts personally identifiable information (PII), and allows you to enhance the accuracy of a transcription by providing custom vocabularies specific to your industries or use case, or by using custom language models.

The second step involves using foundation models (FMs) with Amazon Bedrock to summarize the content, identify topics, and recognize conclusions, extracting valuable insights that can guide strategic decisions and innovations. Automatic generation of new content also adds value, increasing creativity and productivity.

Generative AI is reshaping the way we analyze audio transcripts, enabling you to unlock insights such as customer sentiment, pain points, common themes, avenues for risk mitigation, and more, that were previously obfuscated.

Use case overview

In this post, we discuss three example use cases in detail. The code artifacts are in Python. We used a Jupyter notebook to run the code snippets. You can follow along by creating and running a notebook in Amazon SageMaker Studio.

Audio summarization and insights, and automated generation of new content using Amazon Transcribe and Amazon Bedrock

Through this use case, we demonstrate how to take an existing marketing asset (a video) and create a new blog post to announce the launch of the video, create an abstract, and extract the main topics and the search engine optimization (SEO) keywords present in the post for documenting and categorizing the asset.

Transcribe audio with Amazon Transcribe

In this case, we use an AWS re:Invent 2023 technical talk as a sample. For the purpose of this notebook, we downloaded the MP4 file for the recording and stored it in an Amazon Simple Storage Service (Amazon S3) bucket.

The first step is to transcribe the audio file using Amazon Transcribe:

# Create a Amazon Transcribe transcirption job by specifying the audio/video file's S3 location
import boto3
import time
import random
transcribe = boto3.client('transcribe')
response = transcribe.start_transcription_job(
    TranscriptionJobName=f"podcast-transcription-{int(time.time())}_{random.randint(1000, 9999)}",
    LanguageCode='en-US',
    MediaFormat='mp3',
    Media={
        'MediaFileUri': '<S3 URI of the media file>'
    },
    OutputBucketName='<name of the S3 bucket that will store the output>',
    OutputKey='transcribe_bedrock_blog/output-files/',
    Settings={
        'ShowSpeakerLabels': True,
        'MaxSpeakerLabels': 3
    }
)
max_tries = 60
while max_tries > 0:
    max_tries -= 1
    job = transcribe.get_transcription_job(TranscriptionJobName=response['TranscriptionJob']['TranscriptionJobName'])
    job_status = job["TranscriptionJob"]["TranscriptionJobStatus"]
    if job_status in ["COMPLETED", "FAILED"]:
        if job_status == "COMPLETED":
            print(
                f"Download the transcript fromn"
                f"t{job['TranscriptionJob']['Transcript']['TranscriptFileUri']}."
            )
        break
    else:
        print(f"Waiting for {response['TranscriptionJob']['TranscriptionJobName']}. Current status is {job_status}.")
    time.sleep(10)

The transcription job will take a few minutes to complete.

When the job is complete, you can inspect the transcription output and check the plain text transcript that was generated (the following has been trimmed for brevity):

# Get the Transcribe Output JSON file
s3 = boto3.client('s3')
output_bucket = job['TranscriptionJob']['Transcript']['TranscriptFileUri'].split('https://')[1].split('/',2)[1]
output_file_key = job['TranscriptionJob']['Transcript']['TranscriptFileUri'].split('https://')[1].split('/',2)[2]
s3_response_object = s3.get_object(Bucket=output_bucket, Key=output_file_key)
object_content = s3_response_object['Body'].read()

transcription_output = json.loads(object_content)

# Let's see what we have inside the job output JSON
print(transcription_output['results']['transcripts'][0]['transcript'])

……….Once the alert comes, how do you kind of correlate these alerts, not by just text signals and text passing, but understanding the topology, the infrastructure topology that is supporting that application or that business service. It is the topology that ultimately gives the source of truth when an alert comes, right. That’s what we mean by a correlation that is assisted with topology in in in the thing that ultimately results in finding a probable root cause. And once ……….

After you have validated the existence of the text, you can use Amazon Bedrock to analyze the output:

# Now let's use this transcript to extract insights with a help of a Large Language Model on Amazon Bedrock
# First let's initialize the bedrock runtime client to invoke the model. 
bedrock_runtime = boto3.client('bedrock-runtime')
# Selecting Claude 3 Sonnet
model_id = 'anthropic.claude-3-sonnet-20240229-v1:0'

Using the transcription from the technical talk, we use Amazon Bedrock to call an FM (we use Anthropic’s Claude 3 Sonnet on Amazon Bedrock in this case). You can choose from the language models available on Amazon Bedrock from AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and Amazon.

You can now perform additional tasks.

Extract the main topics with Amazon Bedrock

The following prompt provides instructions to ask the LLM for the main topics in the technical talk:

# Extracting the main topics with Amazon Bedrock
main_topics_prompt = """Based on the contents of <transcript></transcript>, what are the main topics being discussed? Display the topics as a list.

<transcript>
{transcript}
</transcript>
"""

user_message = {"role": "user", "content": main_topics_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
{
"anthropic_version": "bedrock-2023-05-31",
"max_tokens": 1000,
"messages": messages
}
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
main_topics = response_body['content'][0]['text']
print(main_topics)

We have created a prompt that uses prompting best practices for Anthropic’s Claude. In this case, we pass the transcript within the <transcript></transcript> XML tags and ask for main topics discussed. Based on that, we get the following output:

Based on the contents of the transcript, the main topics being discussed are:

  1. AI and Machine Learning in IT Operations
  2. Challenges faced by IT Operations teams
  3. Observable data explosion from cloud, on-premise, and edge infrastructure
  4. The convergence of IT, SRE, and DevOps roles
  5. Discovery to Resolution pipeline for IT Operations
  6. Alert correlation and intelligent alerting using AI/ML
  7. Probable root cause analysis with AI/ML
  8. Automated remediation and runbook automation
  9. Predictive analytics and preventive maintenance using AI/ML
  10. Benefits of AI/ML for IT Operations (proactive operations, reduced human effort, improved service health, cost optimization)

Generate a blog post to announce the video with Amazon Bedrock

The following prompt provides instructions to ask the LLM for the generation of a new blog post announcing the launch of the preceding recording. Just like the preceding prompt, we use prompting best practices for Anthropic’s Claude:

# Generating a blog post to announce the Podcast with Amazon Bedrock
generate_blog_prompt = """The contents of <transcript></transcript> are a transcription of a technical talk video recording. Based on this information write a blogpost that summarizes the technical talk and announces that the recording of this talk is now available for public consumption. Limit the blogpost to 1500 words. Do not include any information that is not mentioned in <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": generate_blog_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
generate_blog = response_body['content'][0]['text']
print(generate_blog)

Based on that, we get the following output:

Here is a draft 1462 word blogpost summarizing the technical talk and announcing the recording is now available:

Title: Revolutionizing IT Operations with AI/ML: A Future-Ready Approach

The world of IT operations is undergoing a seismic shift, driven by the explosion of observable data, the convergence of IT roles, and the advent of hybrid multi-cloud environments. Traditional reactive approaches are no longer sufficient to keep pace with the complexity and scale of modern IT systems. This is where AI and machine learning (ML) come into play, offering a future-ready approach to revolutionize IT operations.

In a recent technical talk, Verma from HPE OpsRamp outlined the drivers behind this AI/ML imperative and demonstrated how this cutting-edge technology is fundamentally transforming IT operations. The recording of this insightful session is now available for public consumption.

The Drivers Behind the AI/ML Imperative

The first driving force behind the adoption of AI/ML in IT operations is the sheer volume of observable data that………………..

Create an abstract with Amazon Bedrock

The following prompt provides instructions to create an abstract summarizing the content of the technical talk:

# Creating an Abstract with Amazon Bedrock
abstract_prompt = """The contents of <transcript></transcript> are a transcription of a technical talk video recording. Based on this information write an abstract for this technical talk in no more than 100 words. Do not include any information that is not mentioned in <transcript></transcript>.

<transcript>
{transcript}
</transcript>
"""

user_message = {"role": "user", "content": abstract_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
{
"anthropic_version": "bedrock-2023-05-31",
"max_tokens": 1000,
"messages": messages
}
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
abstract = response_body['content'][0]['text']
print(abstract)

Based on this, we get the following output:

Here is a 96 word abstract for the technical talk:

This talk discussed how AI and machine learning innovations are transforming IT operations for hybrid multi-cloud workloads. The speakers outlined the challenges of dealing with exploding observable data from modern infrastructure and the convergence of IT roles. They demonstrated how AI/ML techniques like intelligent alerting, alert correlation, probable root cause analysis, and automated remediation can drive more proactive, predictive operations. Key benefits showcased included reduced human effort, improved service health, and cost optimization. The talk featured a demo of the OpsRamp platform leveraging AI/ML models to streamline the discovery-to-resolution pipeline for managing edge-to-cloud environments.

Extract SEO keywords from the generated blog post with Amazon Bedrock

The following prompt provides instructions to extract the main SEO keywords from the generated blog post. Based on online research on how to extract SEO keywords from long-form text, we came up with the following list. This demonstrates how you can empower an LLM like Anthropic’s Claude to follow instructions and best practices for a particular task or domain. Also, the prompt specifies that the output should be in JSON. This is helpful in use cases where you want to programmatically get results from an LLM and therefore require consistent formatting. Based on best practices for Anthropic’s Claude, we use the Assistant message in the messages API to pre-fill the model’s response to have further control on the output format:

# Extracting SEO keywords from the generated blog post
SEO_keywords_prompt = """Extract the most relevant keywords and phrases from the given blog post text present in <blog></blog> that would be valuable for SEO (search engine optimization) based on the instructions present in <instructions></instructions> below. The ideal keywords should capture the main topics, concepts, entities, and high-value terms present in the content. Use JSON format with key "keywords" and value as an array of keywords. Skip the preamble; go straight into the JSON. 

<blog>
{textblog}
</blog>

<instructions>
1. Carefully read through the entire blog post text to understand the main topics, concepts, and ideas covered.
2. Identify important nouns, noun phrases, multi-word phrases, and relevant adjective-noun combinations that relate to the core subject matter of the post.
3. Look for words and phrases that potential searchers might use to find content like this.
4. Prioritize terms that are highly specific and relevant to the blog topic over generic words.
5. Vary the keyword length and include both head terms (shorter, more popular keywords) and long-tail terms (longer, more specific phrases).
6. Aim to extract around 10-20 of the most valuable, high-impact keywords and phrases for SEO.
</instructions>
"""

user_message =  {"role": "user", "content": SEO_keywords_prompt.format(textblog = generate_blog)}
assistant_message = {"role": "assistant", "content": '{"keywords": ['}
messages = [user_message, assistant_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
SEO_keywords = response_body['content'][0]['text']
SEO_keywords = '{"keywords": [' + SEO_keywords
SEO_keywords_json = json.loads(SEO_keywords)
print(SEO_keywords_json)

Based on this, we get the following output:

{‘keywords’: [‘AI-driven IT operations’, ‘machine learning IT operations’, ‘proactive IT operations’, ‘predictive IT operations’, ‘AI for hybrid cloud’, ‘AI for multi-cloud’, ‘FutureOps’, ‘AI-assisted IT operations’, ‘AI-powered event correlation’, ‘intelligent alerting’, ‘automated remediation workflows’, ‘predictive analytics for IT’, ‘AI anomaly detection’, ‘AI root cause analysis’, ‘AI-driven observability’, ‘AI for DevOps’, ‘AI for SRE’, ‘AI IT operations management’]}

For consistent formatting and structured output, you can also use the Converse and ConverseStream APIs in Amazon Bedrock and use the tool calling capabilities of the LLMs that offer it.

Generate a new blog post version emphasizing specific SEO keywords with Amazon Bedrock

The following prompt provides instructions to generate a new blog post emphasizing three keywords or phrases from the ones extracted in the previous step. In this case, we use the first three keywords in the returned keywords list. The instructions specified in the prompt are derived from SEO content best practices:

# Extracting SEO keywords from the generated blog post
rewrite_SEO_keywords_prompt = """Rewrite the given blog post text present in <blog></blog> to improve its ranking for 3 target keywords listed in <keywords></keywords>. Rewrite the blog post, naturally incorporating the target keywords multiple times throughout the text. Keep the rewritten content high-quality, readable, and relevant, while strategically emphasizing the keywords to improve SEO.

<blog>
{textblog}
</blog>

<keywords>
{kw1},
{kw2},
{kw3}
</keywords>
"""

user_message = {"role": "user", "content": rewrite_SEO_keywords_prompt.format(textblog = generate_blog, kw1 = SEO_keywords_json['keywords'][0], kw2 = SEO_keywords_json['keywords'][1], kw3 = SEO_keywords_json['keywords'][2])}
messages = [user_message]

body=json.dumps(
{
"anthropic_version": "bedrock-2023-05-31",
"max_tokens": 1000,
"messages": messages
}
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
rewrite_SEO_blog = response_body['content'][0]['text']
print(rewrite_SEO_blog)

Based on this, we get the following output:

<blog>

Title: Unleash the Power of AI-Driven and Machine Learning IT Operations for Proactive IT Operations

In today’s rapidly evolving digital landscape, traditional IT operations approaches are struggling to keep pace. The explosion of data from cloud workloads, Kubernetes clusters, and distributed edge locations has overwhelmed conventional monitoring tools, leading to fragmented views and reactive operations that respond to issues after they occur. To stay ahead of the curve, organizations must embrace AI-driven and machine learning IT operations, paving the way for proactive IT operations that enhance efficiency and minimize downtime.

At a recent technical talk, Verma and Radha Krishna Kunni from OpsRamp (recently acquired by HPE) delved into the transformative impact of AI and machine learning on IT operations, DevOps, and SRE for hybrid multi-cloud environments. They highlighted the key challenges ops teams face today and introduced the innovative “FutureOps” approach, which leverages AI and machine learning to revolutionize IT operations.

The full video recording of this insightful technical talk is now available [link], providing a comprehensive understanding of…………

Summarize content discussed in a recorded meeting using Amazon Transcribe and Amazon Bedrock

Through this use case, we demonstrate how to take an existing recording from a meeting (we use a recording from an AWS earnings call) to summarize the content discussed, extract the key points, and provide details on the sentiment of the meeting. For additional information on this use case, see Live Meeting Assistant with Amazon Transcribe, Amazon Bedrock, and Amazon Bedrock Knowledge Bases or Amazon Q Business.

Transcribe audio with Amazon Transcribe

In this use case, we use an Amazon 2024 Q1 earnings call as a sample. For the purpose of this notebook, we downloaded the WAV file for the recording and stored in an S3 bucket.

The first step is to transcribe the audio file using Amazon Transcribe:

# Create a Amazon Transcribe transcription job by specifying the audio/video file's S3 location
import boto3
import time
import random
transcribe = boto3.client('transcribe')
response = transcribe.start_transcription_job(
    TranscriptionJobName=f"meeting-transcription-{int(time.time())}_{random.randint(1000, 9999)}",
    LanguageCode='en-US',
    MediaFormat='mp3',
    Media={
        'MediaFileUri': '<S3 URI of the media file>'
    },
    OutputBucketName='<name of the S3 bucket that will store the output>',
    OutputKey='transcribe_bedrock_blog/output-files/',
    Settings={
        'ShowSpeakerLabels': True,
        'MaxSpeakerLabels': 10
    }
)
# Check whether the transcribe job is complete

max_tries = 60
while max_tries > 0:
    max_tries -= 1
    job = transcribe.get_transcription_job(TranscriptionJobName=response['TranscriptionJob']['TranscriptionJobName'])
    job_status = job["TranscriptionJob"]["TranscriptionJobStatus"]
    if job_status in ["COMPLETED", "FAILED"]:
        if job_status == "COMPLETED":
            print(
                f"Download the transcript fromn"
                f"t{job['TranscriptionJob']['Transcript']['TranscriptFileUri']}."
            )
        break
    else:
        print(f"Waiting for {response['TranscriptionJob']['TranscriptionJobName']}. Current status is {job_status}.")
    time.sleep(10)

The transcription job will take a few minutes to complete.

When the job is complete, you can inspect the transcription output and check for the plain text transcript that was generated:

import json
# Get the Transcribe Output JSON file
s3 = boto3.client('s3')
output_bucket = job['TranscriptionJob']['Transcript']['TranscriptFileUri'].split('https://')[1].split('/',2)[1]
output_file_key = job['TranscriptionJob']['Transcript']['TranscriptFileUri'].split('https://')[1].split('/',2)[2]
s3_response_object = s3.get_object(Bucket=output_bucket, Key=output_file_key)
object_content = s3_response_object['Body'].read()

transcription_output = json.loads(object_content)

# Let's see what we have inside the job output JSON
print(transcription_output['results']['transcripts'][0]['transcript'])

Thank you for standing by. Good day, everyone and welcome to the amazon.com first quarter, 2024 financial results teleconference. At this time, all participants are in a listen only mode. After the presentation, we will conduct a question and answer session. Today’s call is being recorded and for opening remarks, I’ll be turning the call over to the Vice President of Investor………

After you have validated the existence of the text, you can use Amazon Bedrock to analyze the output:

# Now let's use this transcript to extract insights with a help of a Large Language Model on Amazon Bedrock
# First let's initialize the bedrock runtime client to invoke the model. 
bedrock_runtime = boto3.client('bedrock-runtime')
# Selecting Claude 3 Sonnet
model_id = 'anthropic.claude-3-sonnet-20240229-v1:0'

Using the transcription from the earnings call recording, we use Amazon Bedrock to call an FM (we use Anthropic’s Claude 3 Sonnet in this case). You can choose from other FMs available on Amazon Bedrock.

You can now perform additional tasks.

Identify the financial ratios highlighted during this earnings call

The following prompt provides instructions to identify financial ratios highlighted during the earnings call and their implications:

# Identify the financial ratios highlighted during this earnings call 
financial_ratios_prompt = """Based on the contents of <transcript></transcript>,identify the financial ratios highlighted during this earnings call and their implications <transcript></transcript> . 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": financial_ratios_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
financial_ratios = response_body['content'][0]['text']
print(financial_ratios)

Based on this, we get the following output:

Based on the earnings call transcript, the following financial ratios and their implications were highlighted:

  1. Operating Income Margin:
    • Amazon reported its highest ever quarterly operating income of $15.3 billion, which was $3.3 billion above the high end of their guidance range. This was driven by strong operational performance across all three reportable segments (North America, International, and AWS) and better-than-expected operating leverage, including lower cost to serve.
    • North America segment operating income was $5 billion with an operating margin of 5.8%, up 460 basis points year-over-year, driven by improvements in cost to serve, including benefits from regionalization efforts, more consolidated customer shipments, and improved leverage.
    • International segment operating income was $903 million with an operating margin of 2.8%, up 710 basis points year-over-year, primarily driven by cost efficiencies through network design enhancements and improved volume leverage in established countries, as well as progress in emerging countries.
    • AWS operating income was $9.4 billion, an increase of $4.3 billion year-over-year, with improved leverage from managing infrastructure and fixed costs while growing at a healthy rate.

Implication: The higher operating income margins across all segments indicate Amazon’s focus on driving efficiencies and improving profitability while continuing to invest in growth opportunities.

  1. Revenue Growth:
    • Worldwide revenue was $143.3 billion, up 13% year-over-year (excluding the impact of foreign exchange).
    • AWS revenue grew 17.2% year-over-year, accelerating from 13.2% in Q4 2023, driven by strong demand for both generative AI and non-generative AI workloads.
    • Advertising revenue grew 24% year-over-year (excluding the impact of foreign exchange), primarily driven by sponsored products and improvements in relevancy and measurement capabilities.

Implication: The strong revenue growth, particularly in AWS and advertising, highlights Amazon’s diversified revenue streams and the growth opportunities in cloud computing and digital advertising.

  1. Capital Expenditures (Capex):
    • Amazon anticipates a meaningful increase in overall capital expenditures in 2024, primarily driven by higher infrastructure Capex for growth in AWS, including generative AI investments.
    • In Q1 2024, Capex was $14 billion, expected to be the lowest quarter of the year.

Implication: The increase in Capex signals Amazon’s confidence in the strong demand for AWS and their commitment to investing in emerging technologies like generative AI to drive future growth.

Overall, the financial ratios and commentary indicate Amazon’s focus on improving profitability, driving operational efficiencies, and investing in growth opportunities, particularly in AWS and generative AI, while maintaining a diversified revenue stream and managing costs effectively.

Identify the speakers from the earnings call with Amazon Bedrock

The following prompt provides instructions to identify the speakers in the meeting from the transcription:

# Identify the speakers from the earnings call 
speakers_prompt = """Based on the contents of <transcript></transcript>,identify the speakers on this earnings call <transcript></transcript> . 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": speakers_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
speakers = response_body['content'][0]['text']
print(speakers)

Based on this, we get the following output:

Based on the transcript, the key speakers on this Amazon earnings call appear to be:

  1. Andy Jassy – CEO of Amazon
  2. Brian Olsavsky – CFO of Amazon
  3. Dave Fildes – Vice President of Investor Relations at Amazon

The call begins with opening remarks from Dave Fildes, followed by prepared statements from Andy Jassy and Brian Olsavsky. They then take questions from analysts, with Andy and Brian providing the responses.

Obtain the challenges or negative areas discussed on the earnings call with Amazon Bedrock

The following prompt provides instructions to obtain the challenges or negative areas discussed from the transcription:

# Obtain the challenges or negative areas discussed on earnings call

challenges_prompt = """Based on the contents of <transcript></transcript>,Obtaining the challenges or negative areas discussed on earnings <transcript></transcript> . 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": challenges_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
challenges = response_body['content'][0]['text']
print(challenges)

Based on this, we get the following output:

Based on the transcript, some of the key challenges or negative areas discussed include:

  1. Foreign exchange headwinds: Amazon faced an unfavorable impact from global currencies weakening against the U.S. dollar in Q1, leading to a $700 million or 50 basis point headwind to revenue compared to guidance.
  2. Increasing capital expenditures: Amazon expects to meaningfully increase its capital expenditures year-over-year in 2024, primarily driven by higher infrastructure spending for AWS growth, including investments in generative AI capabilities.
  3. Consumer spending concerns: Amazon mentioned keeping an eye on consumer spending trends, specifically in Europe, where it appears weaker relative to the U.S.
  4. International segment profitability: While the international segment’s profitability improved, with an operating margin of 2.8%, Amazon acknowledged the need to continue working on cost efficiencies and profitability, particularly in emerging countries.
  5. Cost optimization challenges: Although Amazon believes the majority of cost optimization efforts are behind them, there is still a need to continually streamline processes, optimize inventory placement, and invest in automation to further reduce the cost to serve.

Overall, the challenges centered around foreign exchange impacts, increasing capital intensity for AWS and generative AI investments, consumer demand uncertainties, and ongoing efforts to improve operational efficiencies and international profitability.

Get insights from a call center call between an agent and a customer using Amazon Transcribe and Amazon Bedrock

Through this use case, we demonstrate how to take an existing call recording from a contact center and summarize the content discussed, extract the main topic, key phrases, call reason, customer satisfaction, overall call sentiment, and sentiment about the products and services discussed. For additional details about this use case, see Live call analytics and agent assist for your contact center with Amazon language AI services and Post call analytics for your contact center with Amazon language AI services.

Transcribe audio with Amazon Transcribe

The first step is to transcribe the audio file using Amazon Transcribe. In this case, we use a sample from the Amazon Transcribe Post Call Analytics Solution GitHub repository. For the purpose of this notebook, we downloaded the WAV file and stored it in an S3 bucket.

# Create a Amazon Transcribe transcirption job by specifying the audio/video file's S3 location
import boto3
import time
import random
transcribe = boto3.client('transcribe')
response = transcribe.start_transcription_job(
    TranscriptionJobName=f"call_center-transcription-{int(time.time())}_{random.randint(1000, 9999)}",
    LanguageCode='en-US',
    MediaFormat='mp3',
    Media={
        'MediaFileUri': '<S3 URI of the media file>'
    },
    OutputBucketName='<name of the S3 bucket that will store the output>',
    OutputKey='transcribe_bedrock_blog/output-files/',
    Settings={
        'ShowSpeakerLabels': True,
        'MaxSpeakerLabels': 3
    }
)
max_tries = 60
while max_tries > 0:
    max_tries -= 1
    job = transcribe.get_transcription_job(TranscriptionJobName=response['TranscriptionJob']['TranscriptionJobName'])
    job_status = job["TranscriptionJob"]["TranscriptionJobStatus"]
    if job_status in ["COMPLETED", "FAILED"]:
        if job_status == "COMPLETED":
            print(
                f"Download the transcript fromn"
                f"t{job['TranscriptionJob']['Transcript']['TranscriptFileUri']}."
            )
        break
    else:
        print(f"Waiting for {response['TranscriptionJob']['TranscriptionJobName']}. Current status is {job_status}.")
    time.sleep(10)

The transcription job will take a few minutes to complete.

When the job’s complete, you can inspect the transcription output and check for the plain text transcript that was generated:

import json
# Get the Transcribe Output JSON file
s3 = boto3.client('s3')
output_bucket = job['TranscriptionJob']['Transcript']['TranscriptFileUri'].split('https://')[1].split('/',2)[1]
output_file_key = job['TranscriptionJob']['Transcript']['TranscriptFileUri'].split('https://')[1].split('/',2)[2]
s3_response_object = s3.get_object(Bucket=output_bucket, Key=output_file_key)
object_content = s3_response_object['Body'].read()

transcription_output = json.loads(object_content)

# Let's see what we have inside the job output JSON
print(transcription_output['results']['transcripts'][0]['transcript'])

Thank you for calling Big Jim’s Auto. This is Travis. How can I help you today? Hello, my name is Violet King and I bought a car not too long ago and a light is coming on um a light on the dashboard. And so I was wondering what I should do about that. Ok. It may depend on what kind of light we’re looking at here today, ma’am. Uh Could I get your first and last name spelled out for me so I can just get some information pulled up? Yes. My name is Violet Vviolet. My last name is King King. Ok, I got the call last week. Ok. Uh And what kind of car are we examining today? It’s, it’s a Ford Fusion. It’s 2017, 2017 Ford Fusion. OK. And for verification, ma’am. Do you happen to know the purchase date of the car? Yes, it, it, it was last Tuesday, August 10th. You say the 10th? Ok. And can you describe to me what kind of light we’re looking at? Y yes, it’s uh I, I think it’s a, an oil, an oil light, an oil light? Ok. Ok. And uh just for clarity on my end, ma’am. Um, uh, is this the first call you’ve made regarding this? Yes. Ok. And, uh, this might be kind of a silly question because I know you just got the car. But sometimes they make me ask a silly question about how many miles has the car been driven since you bought it? Oh. I’m not sure. Should I check? Uh, no, that’s ok. I’ll just, I’ll just put in that. We don’t know at this time. It’s ok. Um, so, um, uh, under the warranty we offer, um, we, the, we don’t handle in house oil changes. Um, we basically, when, when someone buys a car from us, the warranty, we have, it, it covers some stuff like, um, weather damage. Um, and, uh, if the engine light comes on, we take a look at that, but the oil change is something that we just don’t have, uh, here at the dealership that’s a little bit outsourced out and they’re pretty backed up right now because a lot of people have been, uh, staying in due to the recent pandemic and now everyone’s just starting to get out and a whole bunch of places are just completely bogged down. So we have a place that we typically outsource to, um, and they’re, they’re pretty reasonable. They’re about, I wanna say somewhere between 25 and $35 to do an oil change. So it’s really not that bad, but they’re a little bit backed up right now from what I’ve heard, I would recommend giving them a call as soon as you can before they close. Ok. What is their number? Uh, give me a second. Let me just rustle through the desk here. See if I can find their information. Uh, ok. Ok. Yes, I’m all right with them one moment, please. Sure. Ok. All their number is 888 333 2222. Ok, and they can fix my car. Yeah, they should be able to handle the oil change. I’m sorry, that’s not something that we cover under the warranty that uh, we have, um, but they should be able to get you settled and, uh, sorted. Ok. Ok. Thank you. Cool. No problem. Have a good one. Thank you. Yup. Bye.

After you have validated the existence of the text, you can use Amazon Bedrock to analyze the output:

# Now let's use this transcript to extract insights with a help of a Large Language Model on Amazon Bedrock
# First let's initialize the bedrock runtime client to invoke the model. 
bedrock_runtime = boto3.client('bedrock-runtime')
# Selecting Claude 3 Sonnet
model_id = 'anthropic.claude-3-sonnet-20240229-v1:0'

Using the transcription from the recorded call center call, we use Amazon Bedrock to call an FM (we use Anthropic’s Claude 3 Sonnet in this case), but you can choose from the other language models available on Amazon Bedrock.

You can now perform additional tasks.

Summarize the call between agent and client with Amazon Bedrock

The following prompt provides instructions to summarize the call discussion from the transcription:

#Summarize the call between agent and customer
summarize_prompt = """Based on the contents of <transcript></transcript>,summarize the call between agent and customer with focus on resolution <transcript></transcript> . 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": summarize_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
summarization = response_body['content'][0]['text']
print(summarization)

We get the following output:

Based on the contents of the transcript, here is a summary of the call between the agent (Travis) and the customer (Violet King) with a focus on resolution:

Violet King called about a light on the dashboard of her recently purchased 2017 Ford Fusion from Big Jim’s Auto. The light appeared to be an oil light. Travis explained that while their warranty covers certain issues like weather damage and engine lights, it does not cover oil changes. He recommended calling an outsourced oil change service that Big Jim’s Auto typically uses, which charges between $25-35 for an oil change.

Travis provided the phone number for the oil change service (888-333-2222) and mentioned that they are currently backed up due to the recent pandemic. He advised Violet to call them as soon as possible before they close to get her oil changed and resolve the issue with the oil light on her dashboard.

The resolution was for Violet to contact the recommended third-party oil change service to have her car’s oil changed, which should address the oil light issue she was experiencing with her newly purchased vehicle.

Extract the main topics with Amazon Bedrock

The following prompt provides instructions to extract the main topics discussed in the conversation from the transcription:

# Extracting the main topics from the conversation
maintopic_prompt = """The contents of <transcript></transcript> are a transcription of a conversation between agent and client. Based on the information, extract the main topics from the conversation  <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": maintopic_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
main_topic = response_body['content'][0]['text']
print(main_topic)

We get the following output:

Based on the conversation transcript, the main topics appear to be:

  1. Dashboard warning light (specifically an oil light) on a recently purchased 2017 Ford Fusion.
  2. Determining if the issue is covered under the warranty provided by the dealership (Big Jim’s Auto).
  3. Recommendation to contact an external auto service provider (phone number provided) for an oil change service, as the dealership does not handle oil changes in-house.
  4. Confirming that the external auto service provider can likely resolve the oil light issue by performing an oil change.

Extract the key phrases with Amazon Bedrock

The following prompt provides instructions to extract the key phrases discussed in the conversation from the transcription:

# Extracting the key phrases with Amazon Bedrock
keyphrase_prompt = """The contents of <transcript></transcript> are a transcription of a conversation between agent and client. Based on the information, extract the key phrases discussed in the conversation <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": keyphrase_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
keyphrases = response_body['content'][0]['text']
print(keyphrases)

We get the following output:

Based on the conversation transcript, here are the key phrases discussed:

  • Oil light
  • 2017 Ford Fusion
  • Purchase date: August 10th
  • First call regarding the issue
  • Car mileage unknown
  • Warranty does not cover oil changes
  • Outsourced oil change service recommended
  • Oil change service contact number: 888-333-2222
  • Oil change service cost: $25 – $35
  • Service is backed up due to the pandemic

Extract the reason why the client called the call center with Amazon Bedrock

The following prompt provides instructions to extract the reason for this client call to the call center from the transcription:

# Extracting the reason why client called the call center
reason_prompt = """The content of <transcript></transcript> is  transcription of a conversation between agent and client. Based on the information, extract the reason why client called the call center <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": reason_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
reason = response_body['content'][0]['text']
print(reason)

We get the following output:

Based on the transcription, the client called the call center because a light (specifically an oil light) was coming on the dashboard of their recently purchased 2017 Ford Fusion car. The client was seeking guidance on what to do about the oil light being on.

Extract the level of customer satisfaction with Amazon Bedrock

The following prompt provides instructions to extract the level of customer satisfaction experienced by the client from the transcription:

# Extracting the level of Customer Satisfaction 
satisfaction_prompt = """The content of <transcript></transcript> is  transcription of a conversation between agent and client. Based on the information, extract the level of customer satisfaction <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": satisfaction_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
csat= response_body['content'][0]['text']
print(csat)

We get the following output:

Based on the transcript, the level of customer satisfaction seems to be moderate.

Evidence:

  1. The agent provided clear explanations regarding the issue with the oil light and why oil changes are not covered under their warranty.
  2. The agent offered a recommendation for an external service provider that could perform the oil change, along with their contact information.
  3. The customer acknowledged the information provided by the agent, indicating some level of satisfaction with the response.

However, there are no explicit statements from the customer expressing high satisfaction or dissatisfaction. The interaction remains polite and resolves the customer’s initial query, but there is no strong indication of exceptional satisfaction or disappointment.

Obtain the overall customer sentiment with Amazon Bedrock

The following prompt provides instructions to obtain the overall customer sentiment from the transcription:

# Extracting the overall customer sentiment.
sentiment_prompt = """The content of <transcript></transcript> is transcription of a conversation between agent and client. Based on the information, what is the overall customer sentiment of the conversation <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": sentiment_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
sentiment = response_body['content'][0]['text']
print(sentiment)

We get the following output:

Based on the conversation transcript, the overall customer sentiment seems neutral to slightly positive. Although the customer, Violet King, was initially concerned about a warning light on her recently purchased car’s dashboard, the agent (Travis) explained the situation clearly and provided a recommendation for getting an oil change from a third-party service provider. The customer acknowledged and accepted the suggestion without expressing significant frustration or dissatisfaction. The conversation ended on a polite note with the customer thanking the agent.

Obtain sentiment about the products or services discussed during the call with Amazon Bedrock

The following prompt provides instructions to obtain sentiment about products and services discussed during the call from the transcription:

## Obtaining sentiment about the products or services discussed.
sentiment_product_prompt = """The content of <transcript></transcript> is transcription of a conversation between agent and client. Based on the information, what is the overall sentiment about the products discussed in the conversation <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": sentiment_product_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
sentiment_product = response_body['content'][0]['text']
print(sentiment_product)

We get the following output:

Based on the conversation transcript, the overall sentiment about the products (the 2017 Ford Fusion car) discussed seems neutral to slightly positive. There are no major complaints or strong negative sentiments expressed about the car itself. The issue raised is regarding an oil light coming on, which seems to be a routine maintenance matter that needs to be addressed by getting an oil change. The agent helpfully provides the customer with information on where to get an oil change done, although mentioning that the warranty does not cover routine oil changes at the dealership. Overall, there is no indication of major dissatisfaction with the car itself from the conversation.

Create an email to send to the client with the call summary

The following prompt provides instructions to create an email to send to the client after the call is completed, summarizing the call:

## Creating an email to send to the client with the call summary
email_prompt = """The content of <transcript></transcript> is transcription of a conversation between agent and client. Based on the information, create an email to send to the client after the call is completed summarizing the call. 
User formal but friendly english style. <transcript></transcript>. 

<transcript>
{transcript}
</transcript>
"""

user_message =  {"role": "user", "content": email_prompt.format(transcript = transcription_output['results']['transcripts'][0]['transcript'])}

messages = [user_message]

body=json.dumps(
    {
        "anthropic_version": "bedrock-2023-05-31",
        "max_tokens": 1000,
        "messages": messages
    }
)
bedrock_response = bedrock_runtime.invoke_model(body=body, modelId=model_id)
response_body = json.loads(bedrock_response.get('body').read())
email_conversation = response_body['content'][0]['text']
print(email_conversation)

We get the following output:

Dear Ms. Violet King,

I hope this email finds you well. I’m writing to summarize our conversation regarding the issue with your 2017 Ford Fusion that you purchased last Tuesday, August 10th.

During our call, you mentioned that an oil light was illuminated on the dashboard of your vehicle. As discussed, our dealership’s warranty does not cover routine maintenance services like oil changes. However, we have a trusted partner that we typically recommend for such services.

The recommended auto service provider is reachable at 888-333-2222. They are experienced in handling oil changes and should be able to assist you with your vehicle’s needs. Please note that they have been experiencing a high volume of requests due to the recent pandemic, so it’s advisable to call them as soon as possible to schedule an appointment.

Conclusion

Using generative AI through Amazon Bedrock to analyze audio transcripts generated by Amazon Transcribe unlocks valuable insights that would otherwise remain hidden within the audio data. By combining the powerful speech-to-text capabilities of Amazon Transcribe with the natural language understanding and generation capabilities of LLMs like those available through Amazon Bedrock, businesses can more efficiently extract key information, generate summaries, identify topics and sentiments, and create new content from their audio and video assets. This approach not only saves time and resources compared to manual transcription and analysis, but also opens up new opportunities for using existing content in innovative ways.

Whether it’s repurposing marketing materials, quickly capturing key points from meetings, or improving customer experience through call center analytics, the combination of Amazon Transcribe and large language models (LLMs) on Amazon Bedrock provides a powerful solution for unlocking the full potential of audio data.As these use cases have demonstrated, this technology can be applied across various domains, from content creation and SEO optimization to business intelligence and customer service. By staying at the forefront of these advancements, organizations can gain a competitive edge by effectively harnessing the wealth of information contained within their audio and video repositories, driving insights, and making more informed decisions.


About the Authors

Ana Maria Echeverri is an AI/ML Worldwide Service Specialist at AWS, focused on driving adoption of generative AI speech analytics use cases. She has worked in the data and AI industry for over 30 years, with over 10 years focused on helping organizations grow their AI maturity and capabilities for successful execution of AI strategies.

Vishesh Jha is a Senior Solutions Architect at AWS. His area of interest lies in generative AI, and he has helped customers and partners get started with NLP using AWS services such as Amazon Bedrock, Amazon Transcribe, and Amazon SageMaker. He is an avid soccer fan, and in his free time enjoys watching and playing the sport. He also loves cooking, gaming, and traveling with his family.

Bala Krishna Jakka is a Technical Account manager at AWS, with a passion for contact center and generative AI technologies. With extensive expertise in helping organizations use cutting-edge solutions, he thrives on staying ahead of the curve in this rapidly evolving field. When not immersed in the realms of AI and customer experience, he finds joy in the game of cricket, showcasing his skills on the pitch. A devoted family man, he cherishes the moments spent with his loved ones, creating lasting memories and finding balance amidst the demands of his professional pursuits

Read More

NVIDIA Advances Robot Learning and Humanoid Development With New AI and Simulation Tools

NVIDIA Advances Robot Learning and Humanoid Development With New AI and Simulation Tools

Robotics developers can greatly accelerate their work on AI-enabled robots, including humanoids, using new AI and simulation tools and workflows that NVIDIA revealed this week at the Conference for Robot Learning (CoRL) in Munich, Germany.

The lineup includes the general availability of the NVIDIA Isaac Lab robot learning framework; six new humanoid robot learning workflows for Project GR00T, an initiative to accelerate humanoid robot development; and new world-model development tools for video data curation and processing, including the NVIDIA Cosmos tokenizer and NVIDIA NeMo Curator for video processing.

The open-source Cosmos tokenizer provides robotics developers superior visual tokenization by breaking down images and videos into high-quality tokens with exceptionally high compression rates. It runs up to 12x faster than current tokenizers, while NeMo Curator provides video processing curation up to 7x faster than unoptimized pipelines.

Also timed with CoRL, NVIDIA presented 23 papers and nine workshops related to robot learning and released training and workflow guides for developers. Further, Hugging Face and NVIDIA announced they’re collaborating to accelerate open-source robotics research with LeRobot, NVIDIA Isaac Lab and NVIDIA Jetson for the developer community.

Accelerating Robot Development With Isaac Lab 

NVIDIA Isaac Lab is an open-source, robot learning framework built on NVIDIA Omniverse, a platform for developing OpenUSD applications for industrial digitalization and physical AI simulation.

Developers can use Isaac Lab to train robot policies at scale. This open-source unified robot learning framework applies to any embodiment — from humanoids to quadrupeds to collaborative robots — to handle increasingly complex movements and interactions.

Leading commercial robot makers, robotics application developers and robotics research entities around the world are adopting Isaac Lab, including 1X, Agility Robotics, The AI Institute, Berkeley Humanoid, Boston Dynamics, Field AI, Fourier, Galbot, Mentee Robotics, Skild AI, Swiss-Mile, Unitree Robotics and XPENG Robotics.

Project GR00T: Foundations for General-Purpose Humanoid Robots 

Building advanced humanoids is extremely difficult, demanding multilayer technological and interdisciplinary approaches to make the robots perceive, move and learn skills effectively for human-robot and robot-environment interactions.

Project GR00T is an initiative to develop accelerated libraries, foundation models and data pipelines to accelerate the global humanoid robot developer ecosystem.

Six new Project GR00T workflows provide humanoid developers with blueprints to realize the most challenging humanoid robot capabilities. They include:

  • GR00T-Gen for building generative AI-powered, OpenUSD-based 3D environments
  • GR00T-Mimic for robot motion and trajectory generation
  • GR00T-Dexterity for robot dexterous manipulation
  • GR00T-Control for whole-body control
  • GR00T-Mobility for robot locomotion and navigation
  • GR00T-Perception for multimodal sensing

“Humanoid robots are the next wave of embodied AI,” said Jim Fan, senior research manager of embodied AI at NVIDIA. “NVIDIA research and engineering teams are collaborating across the company and our developer ecosystem to build Project GR00T to help advance the progress and development of global humanoid robot developers.”

New Development Tools for World Model Builders

Today, robot developers are building world models — AI representations of the world that can predict how objects and environments respond to a robot’s actions. Building these world models is incredibly compute- and data-intensive, with models requiring thousands of hours of real-world, curated image or video data.

NVIDIA Cosmos tokenizers provide efficient, high-quality encoding and decoding to simplify the development of these world models. They set a new standard of minimal distortion and temporal instability, enabling high-quality video and image reconstructions.

Providing high-quality compression and up to 12x faster visual reconstruction, the Cosmos tokenizer paves the path for scalable, robust and efficient development of generative applications across a broad spectrum of visual domains.

1X, a humanoid robot company, has updated the 1X World Model Challenge dataset to use the Cosmos tokenizer.

“NVIDIA Cosmos tokenizer achieves really high temporal and spatial compression of our data while still retaining visual fidelity,” said Eric Jang, vice president of AI at 1X Technologies. “This allows us to train world models with long horizon video generation in an even more compute-efficient manner.”

Other humanoid and general-purpose robot developers, including XPENG Robotics and Hillbot, are developing with the NVIDIA Cosmos tokenizer to manage high-resolution images and videos.

NeMo Curator now includes a video processing pipeline. This enables robot developers to improve their world-model accuracy by processing large-scale text, image and video data.

Curating video data poses challenges due to its massive size, requiring scalable pipelines and efficient orchestration for load balancing across GPUs. Additionally, models for filtering, captioning and embedding need optimization to maximize throughput.

NeMo Curator overcomes these challenges by streamlining data curation with automatic pipeline orchestration, reducing processing time significantly. It supports linear scaling across multi-node, multi-GPU systems, efficiently handling over 100 petabytes of data. This simplifies AI development, reduces costs and accelerates time to market.

Advancing the Robot Learning Community at CoRL

The nearly two dozen research papers the NVIDIA robotics team released with CoRL cover breakthroughs in integrating vision language models for improved environmental understanding and task execution, temporal robot navigation, developing long-horizon planning strategies for complex multistep tasks and using human demonstrations for skill acquisition.

Groundbreaking papers for humanoid robot control and synthetic data generation include SkillGen, a system based on synthetic data generation for training robots with minimal human demonstrations, and HOVER, a robot foundation model for controlling humanoid robot locomotion and manipulation.

NVIDIA researchers will also be participating in nine workshops at the conference. Learn more about the full schedule of events.

Availability

NVIDIA Isaac Lab 1.2 is available now and is open source on GitHub. NVIDIA Cosmos tokenizer is available now on GitHub and Hugging Face. NeMo Curator for video processing will be available at the end of the month.

The new NVIDIA Project GR00T workflows are coming soon to help robot companies build humanoid robot capabilities with greater ease. Read more about the workflows on the NVIDIA Technical Blog.

Researchers and developers learning to use Isaac Lab can now access developer guides and tutorials, including an Isaac Gym to Isaac Lab migration guide.

Discover the latest in robot learning and simulation in an upcoming OpenUSD insider livestream on robot simulation and learning on Nov. 13, and attend the NVIDIA Isaac Lab office hours for hands-on support and insights.

Developers can apply to join the NVIDIA Humanoid Robot Developer Program.

Read More

Hugging Face and NVIDIA to Accelerate Open-Source AI Robotics Research and Development

Hugging Face and NVIDIA to Accelerate Open-Source AI Robotics Research and Development

At the Conference for Robot Learning (CoRL) in Munich, Germany, Hugging Face and NVIDIA announced a collaboration to accelerate robotics research and development by bringing together their open-source robotics communities.

Hugging Face’s LeRobot open AI platform combined with NVIDIA AI, Omniverse and Isaac robotics technology will enable researchers and developers to drive advances across a wide range of industries, including manufacturing, healthcare and logistics.

Open-Source Robotics for the Era of Physical AI

The era of physical AI — robots understanding physical properties of environments — is here, and it’s rapidly transforming the world’s industries.

To drive and sustain this rapid innovation, robotics researchers and developers need access to open-source, extensible frameworks that span the development process of robot training, simulation and inference. With models, datasets and workflows released under shared frameworks, the latest advances are readily available for use without the need to recreate code.

Hugging Face’s leading open AI platform serves more than 5 million machine learning researchers and developers, offering tools and resources to streamline AI development. Hugging Face users can access and fine-tune the latest pretrained models and build AI pipelines on common APIs with over 1.5 million models, datasets and applications freely accessible on the Hugging Face Hub.

LeRobot, developed by Hugging Face, extends the successful paradigms from its  Transformers and Diffusers libraries into the robotics domain. LeRobot offers a comprehensive suite of tools for sharing data collection, model training and simulation environments along with designs for low-cost manipulator kits.

NVIDIA’s AI technology, simulation and open-source robot learning modular framework such as NVIDIA Isaac Lab can accelerate the LeRobot’s data collection, training and verification workflow. Researchers and developers can share their models and datasets built with LeRobot and Isaac Lab, creating a data flywheel for the robotics community.

Scaling Robot Development With Simulation

Developing physical AI is challenging. Unlike language models that use extensive internet text data, physics-based robotics relies on physical interaction data along with vision sensors, which is harder to gather at scale. Collecting real-world robot data for dexterous manipulation across a large number of tasks and environments is time-consuming and labor-intensive.

Making this easier, Isaac Lab, built on NVIDIA Isaac Sim, enables robot training by demonstration or trial-and-error in simulation using  high-fidelity rendering and physics simulation to create realistic synthetic environments and data. By combining GPU-accelerated physics simulations and parallel environment execution, Isaac Lab provides the ability to generate vast amounts of training data — equivalent to thousands of real-world experiences — from a single demonstration.

Generated motion data is then used to train a policy with imitation learning. After successful training and validation in simulation, the policies are deployed on a real robot, where they are further tested and tuned to achieve optimal performance.

This iterative process leverages real-world data’s accuracy and the scalability of simulated synthetic data, ensuring robust and reliable robotic systems.

By sharing these datasets, policies and models on Hugging Face, a robot data flywheel is created that enables developers and researchers to build upon each other’s work, accelerating progress in the field.

“The robotics community thrives when we build together,” said Animesh Garg, assistant professor at Georgia Tech. “By embracing open-source frameworks such as Hugging Face’s LeRobot and NVIDIA Isaac Lab, we accelerate the pace of research and innovation in AI-powered robotics.”

Fostering Collaboration and Community Engagement

The planned collaborative workflow involves collecting data through teleoperation and simulation in Isaac Lab, storing it in the standard LeRobotDataset format. Data generated using GR00T-Mimic, will then be used to train a robot policy with imitation learning, which is subsequently evaluated in simulation. Finally, the validated policy is deployed on real-world robots with NVIDIA Jetson for real-time inference.

The initial steps in this collaboration have already been taken, having shown a physical picking setup with LeRobot software running on NVIDIA Jetson Orin Nano, providing a powerful, compact compute platform for deployment.

“Combining Hugging Face open-source community with NVIDIA’s hardware and Isaac Lab simulation has the potential to accelerate innovation in AI for robotics,” said Remi Cadene, principal research scientist at LeRobot.

This work builds on NVIDIA’s community contributions in generative AI at the edge, supporting the latest open models and libraries, such as Hugging Face Transformers, optimizing inference for large language models (LLMs), small language models (SLMs) and multimodal vision-language models (VLMs), along with VLM’s action-based variants of  vision language action models (VLAs), diffusion policies and speech models — all with strong, community-driven support.

Together, Hugging Face and NVIDIA aim to accelerate the work of the global ecosystem of robotics researchers and developers transforming industries ranging from transportation to manufacturing and logistics.

Learn about NVIDIA’s robotics research papers at CoRL, including VLM integration for better environmental understanding, temporal navigation and long-horizon planning. Check out workshops at CoRL with NVIDIA researchers.

Read More