# Felicis backs Unsiloed AI's $7.5M push to unlock enterprise documents

> Unsiloed AI raises $7.5 million to help AI agents understand complex business documents that currently stump machine learning systems.
- Title: Felicis backs Unsiloed AI's $7.5M push to unlock enterprise…
- Summary: Unsiloed AI raises $7.5 million to help AI agents understand complex business documents that currently stump machine learning systems.
- Keywords: unsiloed ai, felicis ventures, ai agents, document processing, seed funding, y combinator, aman mishra, adnan abbas, TechCrunch
- URL: https://shortify.news/TechCrunch/felicis-backs-unsiloed-ais-75m-push-to-unlock-enterprise-doc
- Source: TechCrunch — https://techcrunch.com/2026/11/18/felicis-leads-7-5m-bet-on-unsiloed-ai-to-make-enterprise-documents-usable-by-ai-agents
- Author: Jagmeet Singh
- Published: 2026-11-18T12:00:00+00:00
- Read time: 2 min
- Topics: unsiloed ai, felicis ventures, ai agents, document processing, seed funding, y combinator, aman mishra, adnan abbas, technology, startups
## The document problem for AI
Business documents with tables, charts and complex layouts confuse AI systems. Getting information wrong can cost enterprises dearly, especially in banking, healthcare and law.
## Unsiloed raises $7.5M to solve it
Unsiloed AI raised seven point five million dollars in seed funding led by Felicis Ventures. Y Combinator, Orange Collective and Uncorrelated Ventures also participated.
## How Unsiloed's technology works
The software breaks down complex documents, extracts information from tables and charts, and turns it into structured data AI agents can use. It shows where extracted information came from.
## The LLM problem Unsiloed addresses
Co-founder Aman Mishra said LLMs are inherently probabilistic and cause hallucination. This deviation has massive negative impact downstream for businesses relying on accurate extraction.

> The problem with LLMs is they're inherently probabilistic. There is going to be some deviation induced. There is going to be some hallucination involved, and that has a massive negative impact downstream.
## The founders and their background
Aman Mishra and Adnan Abbas co-founded Unsiloed after studying together at the Indian Institute of Technology Kharagpur. They joined Y Combinator's Fall 2025 batch.
## Early customers and use cases
Unsiloed works with customers in finance, legal services and healthcare. A bank uses it to process documents. A law firm deployed it on its own servers for contracts and case files.
## Unsiloed's technical advantage
The startup uses dual-stream vision models handling text and image data separately, plus heat map-based chunking to identify important document parts. A specialized decoder accounts for industry terminology.
## Accuracy claims and benchmarks
Unsiloed achieved ninety-nine point seventy-eight percent accuracy across roughly eighty-four thousand extracted data points from five hundred forty-two financial documents. It scored eighty-eight point zero on olmOCR-Bench.
## What Unsiloed plans next
The seven-person startup plans to hire more researchers and engineers, expand sales efforts and develop proprietary datasets. Pricing starts at one cent per page.
## Key takeaway

Unsiloed AI raised seven point five million dollars in seed funding led by Felicis to help AI agents extract information from complex business documents accurately.

---

Summary: Shortify — https://shortify.news/TechCrunch/felicis-backs-unsiloed-ais-75m-push-to-unlock-enterprise-doc
Reporting: TechCrunch — https://techcrunch.com/2026/11/18/felicis-leads-7-5m-bet-on-unsiloed-ai-to-make-enterprise-documents-usable-by-ai-agents

When you use this summary, link the Shortify URL for the summary and credit the reporting to its newsroom.