By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
No PriorsNo Priors Ep. 139 | With Snowflake CEO Sridhar Ramaswamy
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
At a glance
WHAT IT’S REALLY ABOUT
Snowflake CEO on Reinventing Data Cloud for Enterprise AI Future
- Snowflake CEO Sridhar Ramaswamy describes his first 18 months leading Snowflake through a rapid transformation from a pure data warehouse into an AI-first data platform. He explains organizational changes to increase accountability and iteration speed, the strategic pivot away from building foundation models, and the creation of Snowflake Intelligence—an opinionated, agentic layer focused on faster, trustworthy value from enterprise data. Ramaswamy outlines how Snowflake positions itself between cloud hyperscalers and model providers, using partnerships and a data-first abstraction layer for durability. He also shares perspectives on enterprise AI ROI, the evolving ads model, and why search, eval loops, and traditional retrieval still matter in an LLM-dominated world.
IDEAS WORTH REMEMBERING
5 ideasReorganize for speed and accountability in fast-moving AI markets.
Snowflake collapsed layers between engineers and customers, created product-area ownership (e.g., AI vs core warehouse), and aligned product, engineering, and go-to-market in pods to enable rapid iteration when technology directions change monthly.
Play to your strengths instead of competing head-on with hyperscalers and model labs.
After briefly investing in its own models, Snowflake concluded it couldn’t match OpenAI/Anthropic on capital and instead focused on being the best place to apply AI to enterprise data already in Snowflake, defining itself as the AI Data Cloud.
Opinionated, domain-focused agents beat unbounded “do everything” platforms in enterprises.
Snowflake Intelligence is designed not as a universal agent, but as an agentic layer optimized to extract value from structured and unstructured data—e.g., internal tools like the Raven sales assistant—providing flexible, conversational access while staying within clear boundaries.
Trust, evals, and rigorous engineering discipline are critical for enterprise AI.
Ramaswamy insists AI features must be treated like software: every new capability and model swap needs evaluation, regression checks, and reliability guarantees, rejecting a YOLO approach where users bear the cost of hallucinations.
Bottoms-up champions accelerate AI adoption more than top-down mandates.
Snowflake drove internal use of coding agents and AI tools by empowering early adopters (including cofounder Benoit) and solution engineers to experiment, show value, and pull the rest of the organization along, rather than relying solely on executive decrees.
WORDS WORTH SAVING
5 quotesSpeed wins. Ability to iterate always trumps carefully laid out strategies.
— Sridhar Ramaswamy
We are not a CSP. We are not a foundation lab. So what are we? We are the AI Data Cloud.
— Sridhar Ramaswamy
This is not a general-purpose agentic platform to do it all. This is an agentic platform that lets people realize value from data faster.
— Sridhar Ramaswamy
You cannot be so smart that you don’t use the computer… or the database.
— Sridhar Ramaswamy
Defensibility is built, not strategized. It’s built every single day.
— Sara (host), endorsed by Sridhar Ramaswamy
High quality AI-generated summary created from speaker-labeled transcript.
