Home→Startups & Technology→Felicis leads $7.5 million seed round for Unsiloed AI
Startups & Technology

Felicis leads $7.5 million seed round for Unsiloed AI

Complex tables, charts and layouts can make business documents difficult for AI agents to read reliably. Unsiloed AI has raised $7.5 million in seed funding to convert those documents into structured data, with Silicon Valley venture firm Felicis leading the round.

Felicis leads $7.5 million seed round for Unsiloed AI

Y Combinator, Orange Collective and Uncorrelated Ventures also invested, alongside angel investors Aarthi Ramamurthy and Beth Turner. Unsiloed’s software extracts information from documents and records where each item appeared in the original, allowing users to check the results. Co-founder and CEO Aman Mishra told TechCrunch that errors from large language models can have serious downstream effects.

Mishra 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. Unsiloed, headquartered in San Francisco, is working with customers in finance, legal services and healthcare, Mishra said. He identified voice AI startup Bolna AI as a customer; a bank and a law firm are also using the software, but he did not name them.

The startup competes with document-processing firms including Reducto and Unstructured, as well as products from Microsoft, Amazon and general-purpose AI providers. Mishra said Unsiloed uses “dual-stream” vision models to process text and image data separately, plus “heat map-based chunking” to focus on important areas such as figures, numbers and charts. Its decoder also accounts for industry-specific terminology.

In a test of 542 financial documents provided by an enterprise customer, Unsiloed reported 99.78% accuracy across roughly 84,000 extracted data points. The company’s technical brief said the comparison included major AI models, Microsoft’s Azure Document Intelligence and the customer’s in-house tool; the results have not been independently verified. On the public olmOCR-Bench, Unsiloed said it scored 88.0, versus 66.0 for Reducto’s agentic system and 39.9 for Unstructured, across 1,403 PDF pages.

Pricing starts at one cent per page, Mishra said. The seven-person company plans to use the funding to hire researchers and engineers, expand sales and develop proprietary datasets.

Comments (0)

Leave a comment

No comments yet. Be the first!