AI
Unsiloed AI raises $7.5 million seed round, focusing on complex enterprise document parsing
Felicis led Unsiloed AI's $7.5 million seed round, with participation from Y Combinator and others. The company serves financial, legal and healthcare clients, converting enterprise documents containing tables, charts and complex layouts into structured data usable by AI agents, while providing traceability to original sources to reduce business errors caused by large-model hallucinations.
As enterprises use AI agents to automate work, large amounts of information remain locked in complex documents. Unsiloed AI announced the completion of $7.5 million in seed financing, led by Felicis, with participation from Y Combinator, Orange Collective, Uncorrelated Ventures, and angel investors Aarthi Ramamurthy and Beth Turner.
Enterprise documents often contain tables, charts and complex layouts that general-purpose AI systems may misread; in settings such as banks, hospitals and law firms, incorrect information can be costly. Unsiloed's software breaks down complex documents, extracts information from tables, charts and other elements, converts it into structured data that AI agents can use, and marks where the information appears in the original document for easy enterprise verification.
Co-founder and CEO Aman Mishra says large language models are inherently probabilistic, can deviate and hallucinate, and can cause serious negative downstream effects. He and co-founder Adnan Abbas were classmates at the Indian Institute of Technology Kharagpur and participated in Y Combinator's Fall 2025 batch.
The company is headquartered in San Francisco and already has clients across finance, legal services and healthcare. Mishra says a bank is using its technology to process documents, and a law firm has deployed it on its own servers to handle contracts and old case files; voice AI startup Bolna AI is also using it to import knowledge bases for building voice agents.
Facing competitors such as Reducto, Unstructured, Microsoft and Amazon, Unsiloed says its advantage lies in its “dual-stream” vision model: text tokens and precise image tokens are processed separately, combined with heatmap-based chunking that directs more compute toward key areas such as graphics, numbers and charts, and uses industry terminology decoders to distinguish different business contexts.