This repository is designed to get you started with Amazon Bedrock Agents by providing a set of examples that demonstrate how to orchestrate agentic workflows.
- Analyst assistant using Code Interpretation: Create an analyst assistant agent that can generate and execute python code to analyze different data files using the Bedrock Agents code interpretation capabilities
- Agent using Amazon Bedrock Guardrails: Create a banking assistant agent that integrates with an Amazon Bedrock Guardrail with a topic deny for investment advice
- Agent using Amazon Bedrock Knowledge Bases: Create an agent for customer support for solar panel maintenance
- Agent with long term memory: Create an agent for a travel assistant use case that has memory retention capabilities
- Agent using models not yet optimized for Bedrock Agents: Agents examples for models where the pre-processing, orchestration, knowledge base and post-processing prompts have not yet been optimized for Bedrock Agents
- AWS CDK Agent: define and deploy a Bedrock Agent using AWS CDK
- Computer use Agent: Computer use agent demo that provides the critical orchestration layer that transforms computer use from a perception capability into actionable automation
- Custom orchestration Agent: Create advanced agents using the custom orchestration functionality of Bedrock Agent
- Configure an inline agent at runtime: Configure and run agents at runtime with inline agents
- Utilize LangChain Tools with Amazon Bedrock Inline Agents: In this code example we will orchestrate a workflow that utilizes LangChain tools like TavilySearchResults, WikipediaQueryRun, and FileManagementToolkit, along with Amazon Bedrock Inline Agents
- Provide conversation history to Amazon Bedrock Agents: In this module, we will create an Amazon Bedrock Agent and understand how to initialize the Agent with Conversation History
- Agent with metadata filtering: Create an agent with metadata filtering to optimize the retrieval of relevant information from a knowledge base
- Agent using OpenAPI schema: Create an insurance claims assistant agent using an OpenAPI schema file for the action groups definition
- Agents with user confirmation before action execution: Create agents that ask for user confirmation before executing an action from an action group
- Agent connected house: Create an agent conected your house surveillance cameras -- Agents with human_in_the_loop: Create an agent with confirmation and return-of-control capabilities.