nsivabalan commented on issue #19256: URL: https://github.com/apache/hudi/issues/19256#issuecomment-4966116297
Really excited to see this proposal. I like that it's intentionally incremental, modular, and starts with read-only capabilities. That feels like the right balance between delivering something useful early while giving the community room to shape the long-term direction. One thought that stayed with me while reading this is that the opportunity feels much bigger than providing an AI interface to Hudi. Over the years, we've built several foundational capabilities like the timeline, metadata table, indexing, incremental processing, table services, concurrency control, etc. None of these were designed specifically for AI, yet they seem like exactly the kinds of primitives an intelligent agent would naturally rely on. It would be exciting if, over time, this evolves into defining what an open, agent-friendly/agentic lakehouse looks like. Not just answering questions about tables, but helping users understand the state of the system, explain behaviors, recommend optimizations, and eventually assist with safe operational tasks in a transparent and observable manner. I also like that the proposal keeps the core independent and treats the agent layer as an optional capability. That should give us the flexibility to evolve rapidly as the AI ecosystem changes, while keeping Hudi's core architecture clean and focused. Looking forward to seeing this evolve. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
