SF : AI Night at the AWS AI Loft in San Francisco

AI Night at the AWS Gen AI Loft in First Market Tower, hosted by ClickHouse and Docker. Three sessions worth flagging. Sushant Hiray, Senior Director of AI at RingCentral, opened with how a large communications platform is moving agentic AI into production rather than living on demos, including the operational tradeoffs you only see when you are running real volume across a real customer base. Lawrence Pan from Variance walked through running enterprise scale on ClickHouse as a primary data lakehouse, with a much leaner stack than I expected, and made a sharp case that the warehouse boundary should move closer to the AI surface than most teams currently allow. Dustin Healy from ClickHouse closed with MCP agents wired through Librechat and Langfuse, and the observability story finally clicked when I saw it run end to end, with traces, evaluations, and tool calls landing in one place you can actually query. The audience was a strong mix of practitioners working on data and AI infrastructure, and the conversations after the talks were some of the most useful of the week, especially around how to instrument agent runtimes for production debugging. The AWS GenAI Loft as a venue keeps doing something right, where the talks are technical enough to be worth showing up for and the room stays small enough that you actually meet people. Grateful to ClickHouse and Docker for putting it together.