The Fifth Elephant 2025 Annual Conference CfP

Atlan is a control plane for data and AI that helps humans of data and AI to discover, understand, govern and collaborate on their data assets effortlessly.

In this talk, I will share how we are building the next generation of search by integrating AI agents, Retrieval-Augmented Generation (RAG), and vector databases into our platform. Instead of relying solely on traditional keyword search or jumping straight to full semantic search (which requires cataloging the entire data estate), we’ve established a practical middle-ground system.

Our approach embeds descriptive context from metadata categories like users, classifications (tags), business metadata, and glossaries into vector stores such as Turbopuffer. Our AI agents perform hybrid retrieval (vector + BM25), rerank results based on relevance, and use LLMs to generate Elasticsearch DSL and summaries. This enables fast, accurate, and explainable search through a natural, chat-based interface.

Key Takeaways:

• Discover how Atlan has combined vector databases, LLMs, and AI agents to create a robust, hybrid search engine.

• Explore the architecture behind Atlan’s AI-assisted chat interface for search.

• Learn practical strategies to enhance data discovery and governance using AI.

Who should attend?

This session is for software engineers — frontend, backend, platform, and infrastructure — engineering leaders, and business leaders who are looking to leverage AI to redefine search systems. This talk will provide actionable insights and proven strategies to elevate your product’s search capabilities and drive impactful business outcomes.

About Me:

I’m a Software Engineer at Atlan, working on developing AI-native systems that facilitate intelligent, contextual, and conversational search experience for Atlan users. I’ve contributed to Atlan’s official MCP framework and I am a maintainer of Microsoft’s AutoGen framework, where I focus on multi-agent orchestration (ReasoningAgent, SwarmAgent, etc.). My passion lies in building real-world AI applications that simplify workflows and make data more accessible to everyone.