• event: https://lu.ma/geqzcvrc
• slides: https://pitch.com/v/agentic-summit-6vxp8q
• recording: https://youtu.be/hHDZC73HP08?feature=shared&t=9869

a 10-minute deep dive into how we’ve reimagined search at atlan with ai, covering our approach, technical implementation, and a live demo of the production system, presented in-person at Agentic Summit 2025 bangalore. i started off with the current problem with atlan’s search and how our users are interacting with it today in the world of ai. i compared our search with the google in its early days before ai, where users would type some keywords in the search bar and get websites showing up. similarly, at atlan we show data assets (tables, columns, etc..) when user is looking for metadata with a keyword.

once, i set the context of the problem, i introduced how our historical search works, which is powered by elasticsearch and is very keyword driven along with the metadata filters applied. later, showing some patterns on how users are expecting answers not just a list of results for their natural language queries.

i then introduced how we’ve reimagined search for current users, understanding their search patterns and pain points, and designing search experiences that feel natural and intuitive. then i dived into how ai powered search actually works, covering the core mechanics behind our implementation and how we make search intelligent and contextual for users. we also looked at the tech behind the scenes, including the technical architecture, key design decisions, and the infrastructure that powers the entire search experience. finally, i wrapped up with a live demo of our real production system in action, showing an interactive demonstration of ai search capabilities as they work today.

the search is just the beginning for us, we are investing in building ai agents across the product to seamlessly assist our users and perform actions for them.

please check out the above slides for more details and live demos.