Thursday, November 19, 2026
Bringing You the Future of the Internet
Editorial News
Est. 2023 · Dover, Delaware
Technology
Editorial News

Unsiloed AI Raises $7.5M to Make Enterprise Documents Usable by AI Agents

Unsiloed AI raised a $7.5 million seed round led by Felicis, with Y Combinator and other investors participating. The startup says its software extracts information from complex documents and points users to the original passages.
Nov 19, 2026 · 4:43 AM

·

4 min read

Share

A person works at a laptop beside a printed document
A person works at a laptop beside a printed document. Illustrative stock photo via Pexels.

Unsiloed AI has raised $7.5 million in seed funding led by Felicis to help businesses make enterprise documents usable by AI agents. The San Francisco-headquartered startup is developing software that converts complex files into structured data while showing where extracted information appeared in the original document.

The company says the approach is designed to address errors that can arise when AI systems interpret charts, tables and complicated layouts. TechCrunch reported the funding and the startup’s plans to use it to hire researchers and engineers, expand sales and build proprietary datasets.

How Unsiloed AI Aims to Make Enterprise Documents Usable

Business documents can contain tables, charts and layouts that AI systems may misread. The source report says errors could carry serious consequences for organizations such as banks, hospitals and law firms that rely on those files in daily work.

A person turns a page in a stack of documents
A person turns a page in a stack of documents. Illustrative stock photo via Pexels.

Unsiloed’s software breaks down documents, extracts information from different elements and structures the results for AI agents. It also identifies the information’s location in the original file, allowing users to review the extraction against its source.

“The problem with LLMs is they’re inherently probabilistic,” co-founder and CEO Aman Mishra told TechCrunch. “There is going to be some deviation induced. There is going to be some hallucination involved, and that has a massive negative impact downstream.”

Mishra co-founded the company with Adnan Abbas, a former classmate at the Indian Institute of Technology (IIT) Kharagpur. The pair joined Y Combinator’s Fall 2025 batch after developing the technology.

A person points to printed charts beside a laptop
A person points to printed charts beside a laptop. Illustrative stock photo via Pexels.

Make Enterprise Documents Usable With Document-Specific Models

The startup calls its approach “dual-stream” vision models. One stream handles text tokens and another handles image data, Mishra said, while “heat map-based chunking” helps identify important document sections.

The system can direct more computing power toward figures, numbers and charts, according to the report. A specialized decoder also accounts for industry terminology, which can carry different meanings in different business settings.

Mishra said customers in finance, legal services and healthcare are already working with Unsiloed. He told TechCrunch that one bank uses the software to process documents and a law firm runs it on its own servers for contracts and old case files. He did not identify those customers. Mishra named voice AI startup Bolna AI as a customer, saying it uses Unsiloed to ingest knowledge bases for voice agents.

Unsiloed is entering a market that includes document-processing companies Reducto and Unstructured, along with offerings from Microsoft and Amazon. Businesses may also use general-purpose AI models to extract document information. Mishra said Unsiloed’s distinction is its handling of complex documents.

Customer Tests and Planned Expansion

Mishra pointed to a test of 542 financial documents supplied by an enterprise customer. A technical brief shared by the startup reported 99.78% accuracy across roughly 84,000 extracted data points. The results have not been independently verified.

The startup also said it scored 88.0 on olmOCR-Bench, compared with 66.0 for Reducto’s agentic system and 39.9 for Unstructured. The startup conducted those tests on a benchmark covering 1,403 PDF pages, including tables, mathematical expressions, scanned documents and complex layouts.

Mishra said pricing starts at one cent per page. He also said directing computing power to parts needing more attention helps control costs. Unsiloed has seven employees and plans to use the new funding for hiring, sales expansion and proprietary datasets.

The California AI safety lawsuit and Web3 startup incubator coverage provide related reading on AI systems and technology ventures. Readers can also see how finance groups approach digital strategy.

Written by

Block Editorial

administrator
Block Editorial Staff publishes reported coverage and explanatory analysis on cryptocurrency, blockchain, Web3, digital assets and financial technology.
Block Editorial is verified by MuckRack's editorial team. learn more
Contents
  1. How Unsiloed AI Aims to Make Enterprise Documents Usable
  2. Make Enterprise Documents Usable With Document-Specific Models
  3. Customer Tests and Planned Expansion
Read0%

The Block Directory

Continue reading
Newest first
Updated 

Continue reading