Unsiloed AI has raised $7.5 million in seed funding to develop technology that converts complex enterprise documents into structured data for AI agents. The funding round was led by Felicis, with participation from Y Combinator, Orange Collective, Uncorrelated Ventures, and angel investors Aarthi Ramamurthy and Beth Turner.
The San Francisco-based startup aims to solve data extraction challenges in sectors like finance, healthcare, and legal services, where complicated layouts, tables, and charts can cause AI systems to misinterpret information. Unsiloed's software extracts data from these elements and provides citations to the original document to allow for human verification. The company's technology utilizes "dual-stream" vision models that process text and image data separately, alongside a technique called "heat map-based chunking" to prioritize important document sections.
“The problem with LLMs is they’re inherently probabilistic,” Unsiloed co-founder and CEO Aman Mishra said. “There is going to be some deviation induced. There is going to be some hallucination involved, and that has a massive negative impact downstream.”
Mishra and co-founder Adnan Abbas, former classmates at the Indian Institute of Technology (IIT) Kharagpur, joined the Y Combinator Fall 2025 batch. The company reported that its technology is already being used by voice AI startup Bolna AI, as well as unnamed clients in the banking and legal sectors. One law firm has deployed the software on its own servers to manage contracts and historical case files.
In a technical brief, the company stated that its system achieved 99.78% accuracy across 84,000 data points during a test of 542 financial documents, outperforming Microsoft’s Azure Document Intelligence and an in-house tool. The results of this test have not been independently verified. The startup also reported scoring 88.0 on the olmOCR-Bench public benchmark, compared to 66.0 for Reducto and 39.9 for Unstructured.
Unsiloed enters a competitive market featuring specialized document-processing firms like Reducto and Unstructured, as well as large-scale offerings from Amazon and Microsoft. To differentiate its service, the company uses a specialized decoder designed to recognize industry-specific terminology that may change meaning depending on the business context.














