Skip to content
AI.info

The Pulse

Unsiloed AI raises $7.5 million seed round, TechCrunch reports

Felicis led a seed round for Unsiloed AI, which makes software to turn complex documents into data for AI agents, TechCrunch reports. The startup plans to hire and expand sales as it competes with other document-processing providers.

Unsiloed AI raises $7.5 million seed round, TechCrunch reports

Unsiloed AI has raised $7.5 million in seed funding led by Felicis to develop software that makes complex business documents usable by AI agents, TechCrunch reports. 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 co-founded by CEO Aman Mishra and Adnan Abbas, according to the report. The pair were former classmates at the Indian Institute of Technology Kharagpur and joined Y Combinator’s Fall 2025 batch.

Funding plans

Mishra told TechCrunch the company plans to use the funding to hire researchers and engineers, expand sales and develop proprietary datasets. The startup has seven employees.

The software extracts information from tables, charts and other parts of documents and turns it into structured data for AI agents, the outlet reports. It also identifies where extracted information came from in the original document, allowing businesses to check the output.

Reading complex files

According to the report, Unsiloed uses what it calls dual-stream vision models to handle text and image data separately, alongside a technique called heat map-based chunking to identify important parts of a document. The technology directs more computing power toward areas such as figures, numbers and charts.

The startup also uses a decoder that takes industry-specific terminology into account, TechCrunch reports. It is intended to help distinguish terms that can have different meanings in different business contexts.

Unsiloed is working with customers in finance, legal services and healthcare, Mishra told the outlet. He said an unnamed bank uses the technology to process documents and an unnamed law firm has deployed it on its own servers for contracts and old case files.

He named voice AI startup Bolna AI as another customer, saying it uses Unsiloed to ingest knowledge bases for building voice agents. Pricing starts at one cent per page, Mishra told TechCrunch.

Tests and competition

A test using 542 financial documents supplied by an enterprise customer produced 99.78% accuracy across roughly 84,000 extracted data points, according to a technical brief the startup shared with TechCrunch. The results were not independently verified, the outlet reported; the test compared Unsiloed with major AI models, Microsoft’s Azure Document Intelligence and the customer’s in-house extraction tool.

In a separate benchmark post, Unsiloed said its parser scored 88.0 on the deterministic pass rate for olmOCR-Bench, an optical character recognition evaluation published by Allen AI. The company said the benchmark runs 8,413 unit tests across 1,403 PDF pages, including tests of tables, mathematical notation, reading order and old scans.

Its reported results put Nanonets OCR-3 at 87.4 and GPT-5.5 at 84.6. The Nanonets scores came from publicly published results and could not be independently verified, the company said.

Unsiloed faces specialized document-processing startups including Reducto and Unstructured, as well as offerings from Microsoft and Amazon, TechCrunch reports. Businesses also sometimes use general-purpose AI models to extract information from documents.

Sources

  1. TechCrunch
  2. Unsiloed AI

Related stories

Explore

More articles