← back
Building Agents at Cloud Scale — Antje Barth, AWS
Original: Building Agents at Cloud Scale — Antje Barth, AWS
Takeaway
Cloud-scale agents are enabled by model-driven SDKs like Strands plus retrieval over tool catalogs to dodge context-window limits.
Summary
- AWS has 1,000+ GenAI apps in development; Alexa Plus shipped to 600M devices using hundreds of specialized expert systems across 10s of thousands of partner services.
- Amazon Q CLI was built and shipped in 3 weeks using a model-driven approach where developers describe what, not how.
- Open-sourced Strands Agents Python SDK: pre-built tools, multimodal support, Bedrock + Anthropic/Meta/Ollama/OpenAI via LiteLLM, MCP native, A2A coming.
- Trick for 6,000-tool internal agent: store tool descriptions in a knowledge base and use a retrieve tool to surface only the few relevant ones into context.
- Comes with graph/swarm multi-agent workflow primitives and an awslabs/mcp GitHub repo of AWS MCP servers.
strandsawsmcp
Original description
Let's explore practical strategies for building and scaling agents in production. Discover how to move from local MCP implementations to cloud-scale architectures and how engineering teams leverage these patterns to develop sophisticated agent systems. Expect a mix of demos, use case discussions, and a glimpse into the future of agentic services! About Antje Barth Antje Barth is a Principal Developer Advocate at AWS, based in San Francisco. She frequently speaks at AI engineering conferences, events, and meetups, and works closely with product teams to build the future of agentic AI. Antje is also co-author of the O’Reilly books Generative AI on AWS and Data Science on AWS. Recorded at the AI Engineer World's Fair in San Francisco. Stay up to date on our upcoming events and content by joining our newsletter here: https://www.ai.engineer/newsletter