Felicis leads $7.5M bet on Unsiloed AI to make enterprise documents usable by AI agents

Glass Epoch DeskOriginal digestSource: TechCrunch AI

AI-assisted digest, checked against the original source before publishing. How we work

A seven-person startup backed by Felicis and Y Combinator wants to make messy enterprise documents trustworthy enough for AI agents to act on.
Felicis leads $7.5M bet on Unsiloed AI to make enterprise documents usable by AI agents
Photo: Unsplash · free stock license

Unsiloed AI has raised $7.5 million in seed funding led by Felicis to build technology that makes enterprise documents usable by AI agents. Y Combinator, Orange Collective, and Uncorrelated Ventures joined the round, along with angel investors Aarthi Ramamurthy and Beth Turner. The San Francisco-headquartered startup was founded by Aman Mishra and Adnan Abbas, former classmates at IIT Kharagpur who joined Y Combinator's Fall 2025 batch after developing the technology.

The problem Unsiloed targets is specific: business documents are dense with tables, charts, and complicated layouts that AI systems struggle to interpret accurately, and mistakes carry real cost for banks, hospitals, and law firms that depend on those documents daily. Unsiloed's software breaks down complex documents, extracts information from tables and charts, and converts it into structured data AI agents can consume. It also surfaces where each piece of extracted information originated in the original document, so businesses can verify output rather than accept it blindly.

Mishra, Unsiloed's co-founder and CEO, framed the underlying risk in probabilistic terms. "The problem with LLMs is they're inherently probabilistic," he said. "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 company is a seven-person team, per the report's framing of its competitive position.

Early deployments span finance, legal services, and healthcare. Mishra said one bank uses the technology to process documents, while a law firm runs it on its own servers to handle contracts and old case files; neither customer was named. Voice AI startup Bolna AI was identified as a customer, using Unsiloed to ingest knowledge bases for building voice agents.

Under the hood, Unsiloed says its edge comes from "dual-stream" vision models that process text and image data separately, paired with "heat map-based chunking" that surfaces the most important parts of a document. "In one of the streams you have the text tokens, in the other stream you have the exact image tokens," Mishra said, calling it something the team has spent considerable time architecting.

The bet lands in a crowded market. Unsiloed competes with specialized document-processing startups Reducto and Unstructured, cloud offerings from Microsoft and Amazon, and general-purpose AI models that businesses already use for extraction. What remains to be seen is whether a small team's specialized approach can win enterprise trust — particularly regulated buyers — against incumbents with broader distribution, and whether verification features like source-linked extraction become a baseline requirement for agent-facing document pipelines.

keyFacts

whyItMatters

Key facts

  • • Felicis leads $7.5M bet on Unsiloed AI to make enterprise documents usable by AI agents | TechCrunch Felicis leads.
  • • Unsiloed AI has raised $7.5 million in seed funding led by Felicis to tackle that problem, aiming to help.
  • • The technology also shows where the extracted information came from in the original document, so businesses can.
  • • “The problem with LLMs is they’re inherently probabilistic,” said Unsiloed co-founder and CEO Aman.

Why it matters

Takeaway: Felicis leads $7.5M bet on Unsiloed AI to make enterprise documents usable by AI agents | TechCrunch Felicis leads $7.5M bet on Unsiloed AI to make enterprise documents usable….

Spotted an error? Send a correction.

Get the day's AI news in one email

The Daily Epoch · 60-second read · unsubscribe anytime

Related