"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis Snowflake VP of AI Baris Gultekin on Bringing AI to Data, Agent Design, Text-2-SQL, RAG & More
255 snips
Jan 14, 2026 
Baris Gultekin, Vice President of AI at Snowflake, leads the charge in integrating AI with enterprise data while ensuring security and governance. He discusses the evolution from structured analytics to unlocking unstructured data, making AI more accessible without analysts. Baris highlights advancements in text-to-SQL, emphasizes the importance of embedding quality for RAG, and predicts that agents will enhance productivity in product development. His insights on open standards and model choices illuminate the future of AI in transforming data analytics.
AI Snips
Chapters
Transcript
Episode notes
Multiple Model Options And Customer Preferences
- Snowflake offers customers a choice of frontier and open models (OpenAI, Anthropic, Gemini, Meta, Mistral) and also provides its own models like DeepSeek.
- Customer acceptance of model sources varies idiosyncratically, not strictly by industry.
Run Inference Inside The Security Boundary
- Snowflake runs inference "inside the Snowflake security boundary" via partnerships so customer data doesn't leave governed boundaries.
- Models act as sub‑processors with legal guarantees and no persistent state saved, addressing regulated customers' needs.
Clouds And Labs Secure Model IP
- Cloud and model providers enforce strict environments and access controls to protect model IP and weights during inference.
- Snowflake relies on these secured setups and legal relationships to deliver assurances to enterprises.