I want to build some AI Agents connect to my Kafka cluster to analyse some streams. What approach have people taken? The Flink community are marketing this quite a lot, is this a valid solution?
Absolutely Lenses offers a fully functional MCP server that connects AI tools such as AI Agents and LLM chat interfaces up to Lenses and through Lenses to your streaming data in Kafka
You can use the powerful Lenses IAM model to limit what LLMs and agents can see and do with your streaming data.
Lenses MCP server is open source and fully available here:
Check it out and empower your LLMs to connect to your streaming data!
Thanks @Drew_Oetzel!
@Teddy_Kuk Lenses MCP server can be considered your LLM integration for Kafka since MCP is supported in agent frameworks. I’ve used Strands Agents, LangGraph and FastMCP to build agents that analyse streams and I prefer using MCP rather than direct integrations (function calls) because it’s an emerging open standard.
Give it a whirl and please feel free to open a GitHub issue if you have any suggestions for improvements 