The Stack Overflow Podcast Democratizing your data access with AI agents
Sep 23, 2025
Jeff Hollan, Director of Product at Snowflake, dives into how data enhances AI and AI agents. He explains the foundational role of organizational data in providing business context for LLMs. The conversation covers Snowflake's advanced features like vectorized retrieval and the separation of storage and compute for improved performance. Jeff highlights the new Snowflake Marketplace, including the integration of Stack Exchange data, and discusses future challenges in building reliable AI agents. It's a fascinating glimpse into the future of data-driven AI.
AI Snips
Chapters
Transcript
Episode notes
When Generic LLMs Hit Their Limits
- Jeff Hollan recounts using general LLMs for personal tasks but finding them unhelpful for Snowflake-specific product questions.
- He emphasizes that unique business context lives in an organization's data and must be connected to AI for useful answers.
Four Core Building Blocks For Enterprise AI
- Effective AI requires fast vector lookups, retrieval-augmented generation, SQL querying, and governance controls.
- Snowflake bundles these building blocks to supply relevant snippets, live queries, and access controls for enterprise AI.
Query Engine Is Central To The Platform
- Snowflake provides the query engine for most customers and can integrate data from various storage formats.
- This lets agents query near-real-time production or analytics data without leaving the platform.