Unsiloed AI has raised $7.5 million in seed funding, led by Silicon Valley venture capital firm Felicis, to address the challenge of making complex enterprise documents usable by AI agents. The funding round also saw participation from Y Combinator, Orange Collective, Uncorrelated Ventures, and angel investors Aarthi Ramamurthy and Beth Turner.
The San Francisco-headquartered startup aims to develop software that can break down complex documents, extract information from elements like tables and charts, and convert it into structured data for AI agents. This technology also indicates the origin of extracted information within the original document, allowing businesses to verify results.
Unsiloed AI co-founder and CEO Aman Mishra stated that the company's approach uses "dual-stream" vision models, which process text and image data separately, and "heat map-based chunking" to identify important document sections. The technology also directs more computing power to areas such as figures and charts and uses a specialized decoder for industry-specific terminology.
Mishra reported that Unsiloed AI is currently working with customers in finance, legal services, and healthcare. One bank is reportedly using its technology for document processing, and a law firm has deployed it for contracts and old case files. Voice AI startup Bolna AI is also named as a customer, using Unsiloed to ingest knowledge bases for building voice agents.
A technical brief shared by Unsiloed AI claims 99.78% accuracy across approximately 84,000 extracted data points in a test involving 542 financial documents from an enterprise customer. These results, which compared Unsiloed with major AI models, Microsoft’s Azure Document Intelligence, and the customer’s in-house tool, have not been independently verified. Unsiloed also claims to have outperformed competing tools on the olmOCR-Bench public benchmark.
Unsiloed AI, which currently has a headcount of seven, plans to use the new funding to hire more researchers and engineers, expand sales efforts, and develop proprietary datasets to improve its technology. Pricing for Unsiloed's services reportedly starts at one cent per page.