The article describes a solution built with Amazon Bedrock AgentCore that automates the creation of client PowerPoint presentations for a financial services team. The workflow replaces manual reading of hundreds of pages, data extraction, slide structuring, and visual generation with an AI agent system that completes the task in under 90 seconds in internal testing. The system deploys a single supervisor agent on AgentCore runtime, which selects between document generation and summarization tools based on user requests. The agent uses the Strands Agents framework to orchestrate specialized tools: Anthropic Claude Sonnet 4.6 for content creation, Amazon Titan Text Embeddings v2 for vector search, and Stability AI Stable Image Inpaint for slide images. A Retrieval Augmented Generation pipeline leverages Amazon OpenSearch Serverless for vector storage and semantic search across multimodal content, and Amazon Bedrock Knowledge Bases to ground responses in enterprise data, reducing hallucinations. The solution exposes a REST API built with .NET 8 minimal APIs, enabling integration with existing enterprise applications via standard HTTP calls. AWS Lambda provides serverless compute, including containerized functions for PowerPoint generation. Security is enforced through Amazon VPC isolation, AWS KMS encryption, and IAM least‑privilege access control. Custom content safety guardrails are created with AWS CDK and applied at model invocation to meet compliance requirements. The architecture includes API Gateway routing to a .NET Chat API on Lambda, which maintains conversation state in DynamoDB and invokes the supervisor agent. Generated documents are stored in S3 and returned as presigned URLs. The deployment process involves building artifacts, deploying infrastructure stacks with AWS CDK, and deploying the supervisor agent container to AgentCore runtime. The article notes that in internal testing with a 90‑page Amazon 2024 Annual Report PDF, the solution produced a presentation in under 90 seconds, illustrating significant time savings for the team.
Generate presentations with Amazon Bedrock AgentCore
A financial services team uses Amazon Bedrock AgentCore to automate the creation of client PowerPoint presentations from quarterly reports, compliance documents, and market analyses. The system deploys a supervisor agent on AgentCore runtime that selects between document generation and summarization tools, using Anthropic Claude Sonnet 4.6 for content, Amazon Titan Text Embeddings v2 for vector search, and Stability AI Stable Image Inpaint for slide images. The solution includes a .NET 8 REST API, AWS Lambda functions, Amazon OpenSearch Serverless for semantic search, and security via VPC, KMS, and IAM. In internal testing, the workflow completed in under 90 seconds for a 90‑page report.
Why it matters
Demonstrates how Bedrock AgentCore can drastically reduce manual content creation while maintaining data grounding, a key challenge in enterprise AI deployments.