7-person Unsiloed AI raises $7.5M seed led by Felicis to make documents queryable for AI agents

rohanpaul_ai · x · 2026-11-18

Unsiloed AI, a 7-person startup, raised a $7.5M seed led by Felicis Ventures to build the document layer for AI: converting PDFs, scans and spreadsheets into JSON and Markdown that LLMs and agents can reason over, using proprietary dual-stream vision models — targeting finance, law and healthcare where records still live in PDFs.

The product offers parse, extract and split capabilities with configurable schemas, prompts and confidence thresholds, deployed managed or in a VPC/on-prem. Notably, Unsiloed sells directly to AI agents: its agent storefront publishes prices and payment instructions via llms.txt, letting agents purchase autonomously or with human approval without calling API endpoints. The pain point: a single misread table cell can poison downstream agent reasoning.

Related event: 7-person startup Unsiloed AI raises $7.5M seed round(2 posts)→

Original post →

More from Venture

Venture channel →