Bedrock Agent FAQs

Question 1

How does Bedrock Agent manage the LLM's memory and context?

Accepted Answer

It uses Lambda-hosted MCP (Model Context Protocol) servers for real-time memory and context management, which are backed by an S3 Vector Store for persistent semantic memory and embeddings, allowing for dynamic recall and storage of information.

Question 2

Is the Bedrock Agent architecture serverless and secure?

Accepted Answer

Yes, the entire architecture is fully serverless, providing scalability and cost-efficiency. All API Gateway endpoints for both the agent and MCP server are secured with API key authentication for controlled access.

Question 3

What is Bedrock Agent?

Accepted Answer

Bedrock Agent is a fully serverless solution designed to orchestrate and manage a Bedrock LLM agent. It enables the agent to dynamically store and recall information, ensuring highly relevant and context-aware responses.

Question 4

What AWS services are used by Bedrock Agent?

Accepted Answer

Bedrock Agent leverages AWS Bedrock for language models, AWS Lambda for serverless compute (hosting both the agent and MCP server), S3 Vector Store for persistent memory, and API Gateway for secure API endpoints.

Question 5

How is the Bedrock Agent's infrastructure deployed?

Accepted Answer

The complete infrastructure, including all necessary AWS services like Lambda functions, API Gateway, and S3 Vector stores, is orchestrated and deployed using Python AWS CDK (Cloud Development Kit), enabling infrastructure as code.

Bedrock Agent

About

Key Features

Use Cases