Databricks releases AI Extract for processing 500+ page documents and complex nested schemas
matei_zaharia · x · 2026-08-20
Databricks released AI Extract to tackle complex document processing pain points:
- Processing 500+ page documents exceeding 1M tokens
- Handling large, nested schemas with 1k+ objects
- Managing complex schemas requiring frontier reasoning
Technical Implementation:
- In-house custom model trained on complex real-world examples
- Custom agent harness designed for long documents that decomposes, executes in parallel, and reconciles output
This reflects their methodology: identify key workloads, build in-house models + harnesses, and ship via close collaboration.
More from coding & agent
- Technical discussion: Ditching ORM mapping in the Agent era — JamesTimmins · 2026-08-20
- Deep Agent + Stagehand: Browser Automation with Minimal Code — LangChain · 2026-08-20
- "No ORMs in the agent era": devs debate raw SQL vs object-relational mapping — shakoistsLog · 2026-08-20
- Claude Trading Skills open-sourced: Supports market analysis, risk management, and strategy development — tom_doerr · 2026-08-20
- Simon Willison: LOC metrics matter with agents, but conceptual integrity is harder to keep — Simon Willison · 2026-08-20
- Gemini Notebooks Integrates VM, Antigravity Coding Agent, and Skills Suite — AI_Andrew · 2026-08-20