Super Data Science: ML & AI Podcast with Jon Krohn 938: Frontier AI Agents for Data Science, with Sphinx’s Rohan Kodialam
55 snips
Nov 7, 2025 
Rohan Kodialam, Cofounder and CEO of Sphinx, shares insights on revolutionizing data science with frontier AI. He explains the crucial differences between data work and traditional coding, emphasizing the need for intuition about data as its own modality. Rohan discusses how Sphinx enhances data analysis workflows, making teams five times more productive and allowing for the monetization of previously untapped data. He also highlights the user-friendly, natural-language interface and how Sphinx simplifies using AI for data teams.
AI Snips
Chapters
Books
Transcript
Episode notes
Data As Its Own Modality
- Data science requires intuition about the data itself, not just software engineering skills.
- Sphinx builds an AI layer that treats data as its own modality to enable agentic data work.
Coding-Only Agents Fail On Real Data
- Treating data tasks as pure coding leads to brittle, often incorrect results when data deviates from ideal conditions.
- Humans inspect and interpret data; Sphinx aims to replicate that contextual reasoning for AI.
Visual Intuition Beats Raw Text For Data
- LLMs reading raw numeric tables lack the visual intuition humans get from plots and charts.
- Sphinx encodes data into representations AI can reason over, bridging that gap.
