---
format: "aidr-story-markdown/v1"
id: "0aaf4c6d23854c12d5262c1b164b293f0befa7bfaf7e46cba5b1e9d8ec402013"
canonical_url: "https://aidr.today/0aaf4c6d?lang=en"
title: "Unsiloed AI Raises $7.5M Seed Led by Felicis to Make Documents Queryable for AI"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-18T15:22:08.000Z"
category: "Data"
topics: ["funding","data-engineering"]
source_urls: ["https://marketbrief.now/startups/unsiloed-ai-raises-75m-seed-led-by-felicis-to-make-documents-queryable-f-d768c24d","https://huggingnews.com/startups/unsiloed-ai-raises-75m-seed-led-by-felicis-to-make-documents-queryable-f-d768c24d"]
summary: "Unsiloed AI, a 7-person Y Combinator startup, raised $7.5M in seed funding led by Felicis to build a \"computable layer\" that allows AI agents to query documents. The technology uses computer vision to convert PDFs, scans, and spreadsheets into structured data, targeting industries such as finance, law, and healthcare where records are not typically stored in databases. The company's dual-stream models process document content and layout in parallel, linking every extracted value to its original source for verification. Unsiloed AI reports 99.78% accuracy across 84,000 data points from 542 financial documents, though these figures are not independently verified. The system can be deployed air-gapped on a customer's own servers."
---

# Unsiloed AI Raises $7\.5M Seed Led by Felicis to Make Documents Queryable for AI

> [Open the canonical story](<https://aidr.today/0aaf4c6d?lang=en>)

**Published:** 2026-11-18T15:22:08.000Z
**Category:** Data
**Topics:** funding, data\-engineering

## Summary

Unsiloed AI, a 7\-person Y Combinator startup, raised $7\.5M in seed funding led by Felicis to build a "computable layer" that allows AI agents to query documents\. The technology uses computer vision to convert PDFs, scans, and spreadsheets into structured data, targeting industries such as finance, law, and healthcare where records are not typically stored in databases\. The company's dual\-stream models process document content and layout in parallel, linking every extracted value to its original source for verification\. Unsiloed AI reports 99\.78% accuracy across 84,000 data points from 542 financial documents, though these figures are not independently verified\. The system can be deployed air\-gapped on a customer's own servers\.

## Sources

- [Story source](<https://marketbrief.now/startups/unsiloed-ai-raises-75m-seed-led-by-felicis-to-make-documents-queryable-f-d768c24d>)
- [Story source](<https://huggingnews.com/startups/unsiloed-ai-raises-75m-seed-led-by-felicis-to-make-documents-queryable-f-d768c24d>)

