Felicis Seeds Unsiloed AI with $7.5M for Doc Parsing
Original: Felicis leads $7.5M bet on Unsiloed AI to make enterprise documents usable by AI agents
Why This Matters
Document parsing is a key bottleneck for enterprise AI agent adoption at scale.
Unsiloed AI raised $7.5M seed funding led by Felicis to help AI agents parse complex enterprise documents—tables, charts, layouts—with source attribution. Y Combinator, Orange Collective, and Uncorrelated Ventures also participated.
Enterprise AI agents have a document problem: PDFs full of tables, charts, and messy layouts confuse most AI systems, and errors in finance, legal, or healthcare contexts can be expensive. Unsiloed AI, a San Francisco startup, raised $7.5M in seed funding led by Felicis to fix that. Y Combinator (the company graduated from YC's Fall 2025 batch), Orange Collective, and Uncorrelated Ventures joined the round, alongside angel investors Aarthi Ramamurthy and Beth Turner.
The startup's core tech extracts structured data from complex documents and—critically—traces each extracted item back to its source location, so human reviewers can verify outputs rather than blindly trust AI. CEO Aman Mishra, who co-founded the company with IIT Kharagpur classmate Adnan Abbas, is blunt about the motivation: "LLMs are inherently probabilistic. There is going to be some hallucination involved, and that has a massive negative impact downstream."
Under the hood, Unsiloed uses what it calls "dual-stream" vision models that handle text and image data separately, plus a "heat map-based chunking" technique to identify the most relevant document sections. Customers already include unnamed banking and legal clients, plus voice AI startup Bolna AI, which uses the platform to ingest knowledge bases for voice agents. Competitors include Reducto, Unstructured, and document tools from Microsoft and Amazon.