Aman: Unsiloed Raises a $7.5 Million Seed Led by Felicis
We’ve raised a $7.5M seed round led by Felicis so AI can read documents as easily as it reads databases. Finance, law and healthcare run on documents, and not databases like any other industry. Soon AI agents will read and write more documents than humans, and they’ll need to query them the way software queries tables. We’re building the computable layer that makes this possible. To do this at scale, we've brought together ML researchers and engineers from Google DeepMind, Cognition, and quant trading firms. We use computer vision and dual-stream models to turn complex multimodal documents into formats LLMs and AI agents can understand and reason over. If your team wants to turn messy documents into reliable, AI-ready data, book a demo:
Context
A new document AI startup is raising a sizable seed to make contracts and records as queryable as databases, a bottleneck for agents in law, finance, and healthcare.
Aman, posting from the Unsiloed account, said the company has raised a $7.5 million seed round led by Felicis. He described Unsiloed as a computable layer that turns complex documents into formats models and agents can query, and said the team includes people from Google DeepMind, Cognition, and quant trading firms. The post links to a demo booking page and does not name a valuation or other investors.
Deep dive
Aman announced a $7.5 million seed round led by Felicis for Unsiloed, saying the aim is to let AI read documents as easily as it reads databases. He argued that finance, law, and healthcare still run on documents rather than tables, and that agents will soon read and write more documents than people, so they will need a way to query those files the way software queries tables.
The product claim in the post is that Unsiloed is building that computable layer, using computer vision and dual stream models to turn complex multimodal documents into formats that language models and agents can understand and reason over. Those method names and the promise of reliable, AI ready data are the company's own description, not an independent evaluation.
The post says the team brings together machine learning researchers and engineers from Google DeepMind, Cognition, and quant trading firms. It does not name the people, give a valuation, or list other investors besides Felicis as the lead. A demo booking link is included.
This is a first hand funding announcement from the founder account. It is a seed round, not a major lab release or a large late stage deal, and it should be read as the company's launch pitch plus a confirmed Felicis led raise of $7.5 million.
What to expect next
A fuller round post or a product page should add the valuation, any other investors, and who is already using the system. Watch whether Unsiloed publishes benchmarks on messy contracts or medical records against other document tools. If a later note names the DeepMind or Cognition hires, that can be added here. Until customers test the query layer, the technical claims stay the company's own.
