What happened
Unsiloed AI announced a $7.5 million seed financing round led by Felicis, with participation from Y Combinator, Orange Collective, Uncorrelated Ventures and angels Aarthi Ramamurthy and Beth Turner. The San Francisco startup, founded by IIT Kharagpur batchmates Aman Mishra (CEO) and Adnan Abbas (CTO), builds APIs that convert PDFs, Word files, scanned forms, tables, charts and images into structured Markdown or JSON that large language models and AI agents can consume. The company said the capital will fund further development of its dual‑stream vision architecture, hiring of researchers and engineers, expansion of sales, and creation of proprietary datasets. Pricing is advertised at one cent per page, with private‑cloud, on‑premises and air‑gapped deployment options for enterprises handling sensitive data.
Unsiloed AI closed a $7.5 million seed round, with Felicis Ventures as lead investor. The round also included Y Combinator, Orange Collective, Uncorrelated Ventures and two angel investors. The company did not disclose its post‑money valuation.
The startup’s core product is an API suite that ingests a wide range of document types—PDFs, Word files, scanned forms, tables, charts, and images—and outputs structured data in Markdown or JSON. This output is designed for direct consumption by large language models (LLMs) and autonomous AI agents.
Unsiloed’s architecture relies on a dual‑stream vision model that processes content and visual layout separately, combined with heat‑map based chunking to preserve relational context such as merged cells or pivot numbers. Industry‑specific decoders tailor extraction to sector terminology.
The company reported internal test results showing 99.78 % accuracy across roughly 84,000 data points from 542 financial documents, outperforming Microsoft Azure Document Intelligence and an in‑house tool. On the public olmOCR‑Bench , Unsiloed scored 88.0, ranking first among 19 evaluated systems.
Unsiloed plans to use the new capital to advance its document models, expand its engineering team, hire researchers, grow its sales organization, and build proprietary datasets. The startup also aims to broaden cloud, private‑cloud, on‑premises and air‑gapped deployment options for enterprise customers.
Source details: indianstartupnews.com
Why it matters
Unsiloed AI’s platform tackles a core bottleneck for enterprise AI adoption: the difficulty of extracting reliable, structured data from heterogeneous, often unstructured documents. By delivering deterministic, audit‑able outputs that preserve both content and visual layout, the startup promises to make documents “as computable as databases,” a claim that could reduce the months‑long engineering effort many AI teams currently spend building custom ingestion pipelines. The company’s reported 99.78 % accuracy on a 542‑document financial test set and top‑ranking performance on the public olmOCR‑Bench suggest a competitive edge over general‑purpose services such as Microsoft Azure Document Intelligence and other specialist vendors. If the technology scales, it could accelerate AI‑driven automation across regulated sectors—finance, insurance, healthcare, legal—where data privacy and compliance demand on‑premises processing. The seed round also signals continued investor confidence in niche AI infrastructure that complements large‑model providers rather than competing directly with them.
Document ingestion remains a major friction point for enterprises seeking to leverage LLMs; existing solutions are either probabilistic, lack auditability, or require extensive custom engineering. Unsiloed’s deterministic approach and provenance tracking could dramatically reduce integration time and risk.
The reported accuracy and performance suggest the technology may be ready for production use in high‑stakes domains such as finance, insurance, healthcare, and legal, where errors can have regulatory or financial consequences.
Pricing at one cent per page, combined with flexible deployment models, positions Unsiloed as a cost‑effective alternative to cloud‑only services, especially for organizations with strict data‑privacy mandates.
The seed funding underscores investor belief that specialized AI infrastructure—particularly tools that bridge raw data and LLMs—will be a growth market as enterprises scale AI initiatives.
If Unsiloed can sustain its performance claims in independent evaluations, it could set a new standard for document‑AI, prompting larger cloud providers to improve their own offerings or seek partnerships.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').Which component of an AI application is the machine-learning model itself?
What to watch next
Key indicators to monitor include Unsiloed AI’s ability to convert its performance claims into verifiable, third‑party benchmarks, the uptake of its on‑premises and air‑gapped solutions among Fortune 150 firms, and any strategic partnerships or follow‑on funding that could broaden its market reach. Competitor activity, especially from cloud giants expanding document‑AI services, will also shape the startup’s positioning. Finally, the rollout of its pricing model at one cent per page will reveal whether cost‑effectiveness drives broader enterprise adoption.
Third‑party validation of Unsiloed’s accuracy claims through independent benchmarks or customer case studies.
Adoption metrics from enterprise customers, especially in regulated industries, and any disclosed revenue or contract values.
Potential strategic alliances with large cloud providers or AI platform vendors that could extend the startup’s reach.
Future funding rounds that may signal scaling ambitions or valuation trends for niche AI infrastructure startups.
Competitive moves from Microsoft, Amazon, and other specialist document‑AI firms that could affect market dynamics.