Felicis Leads $7.5M Seed Funding for Unsiloed AI to Help AI Agents Read Complex Enterprise Documents
· GlobeWire Daily Editorial Desk · Source: techcrunch.com

Image: Jagmeet Singh · Photo source · Resized and letterboxed to 16:9; no AI generation
Unsiloed AI has secured $7.5 million in seed funding led by Felicis to help enterprises make complex documents more accessible to AI agents. The funding round included participation from Y Combinator, Orange Collective, Uncorrelated Ventures, and angel investors Aarthi Ramamurthy and Beth Turner.
Business documents often contain intricate tables, charts, and complex layouts that standard AI systems can misinterpret. According to co-founder and CEO Aman Mishra, errors can cause severe downstream issues for industries like finance, healthcare, and law. Unsiloed aims to address this by converting complex files into structured data and providing source citations within the original document for verification.
Technology and Claims
- Dual-Stream Vision Models: Processes text tokens and image tokens separately.
- Heat Map-Based Chunking: Identifies and directs computing power toward critical document sections.
- Industry Decoder: Accounts for specialized terminology across different business contexts.
Mishra reported that a test involving 542 financial documents from an enterprise customer yielded 99.78% accuracy across approximately 84,000 extracted data points, comparing the startup against major AI models and Microsoft’s Azure Document Intelligence. Furthermore, Unsiloed reported scoring 88.0 on the public olmOCR-Bench benchmark by Allen AI, compared to 66.0 for Reducto and 39.9 for Unstructured. These tests and benchmark results have not been independently verified.
Customers and Competition
Mishra stated that the startup is working with customers in finance, legal services, and healthcare, including an unnamed bank and law firm, as well as voice AI startup Bolna AI. Unsiloed enters a crowded market featuring competitors such as Reducto, Unstructured, Microsoft, Amazon, and general-purpose AI models.
Practical Implications and Verification
The seven-person, San Francisco-headquartered startup plans to use the seed capital to hire researchers and engineers, expand sales, and develop proprietary datasets. Pricing starts at one cent per page. Enterprises evaluating document-processing tools can verify Unsiloed's offerings by reviewing technical briefs directly with the company and conducting benchmark tests on internal document sets.
AI-assisted reporting; source claims are attributed and have not been independently verified.



