The Data Exchange with Ben Lorica The AI Revolution Finally Comes to Structured Data
15 snips
Dec 4, 2025 
Jure Leskovec, a Stanford professor and co-founder of Kumo.ai, dives into the transformative power of relational foundation models for structured enterprise data. He challenges the current limitations of AI in handling relational data, emphasizing the shortcomings of treating tabular data as text. Jure outlines Kumo’s rapid predictive SQL-like language, innovative graph representations, and the model's ability to handle messy data effectively. He also discusses real-world successes like DoorDash's significant improvements and the potential applications of these models across various industries.
AI Snips
Chapters
Transcript
Episode notes
Structured Data Holds The Real Business Signal
- Enterprises' most valuable data is structured relational data in warehouses, not text.
- Kumo targets AI predictions directly on this structured ground truth without traditional pipelines.
Traditional ML Pipelines Are Costly And Fragile
- Traditional ML workflows are slow, manual, and costly with long feature pipelines.
- Many models never reach production and demand ongoing maintenance.
Relational Foundation Models Run On Your Schema
- A relational foundation model is a frozen pretrained model that operates over a connected schema of tables.
- It returns accurate predictions in real time and can be fine-tuned for higher business-critical accuracy.