AiGpu

Industry·

Unsiloed AI Secures $7.5M Seed to Turn Complex Documents into AI‑Ready Data

Felicis leads a $7.5 million seed round for Unsiloed AI, a startup that extracts structured data from tables, charts, and layouts so enterprise AI agents can rely on accurate document intelligence.

Unsiloed AI co‑founders Aman Mishra and Adnan Abbas

Unsiloed AI has closed a $7.5 million seed round led by Felicis, with participation from Y Combinator, Orange Collective, Uncorrelated Ventures and notable angels. The San Francisco‑based team, founded by IIT Kharagpur alumni Aman Mishra and Adnan Abbas, is targeting the persistent bottleneck of turning dense, layout‑heavy business documents into clean, machine‑readable data.

The company’s dual‑stream vision architecture processes text tokens and image tokens in parallel, while a heat‑map‑guided chunking mechanism directs compute toward high‑value regions such as tables, figures and numeric fields. A domain‑aware decoder further disambiguates terminology that varies across finance, legal and healthcare contexts.

Why it matters for GPU / AI infrastructure

Accurate document understanding reduces the need for repeated large‑model inference passes, lowering GPU utilisation and latency for downstream agents. By delivering verified, citation‑linked extractions, Unsiloed lets enterprises run lighter, more deterministic workloads on their AI clusters — a direct win for any GPU‑cloud provider focused on cost‑efficient inference.

Early customers include a major bank, a law firm running the stack on‑premises, and voice‑AI startup Bolna AI, signalling strong demand for reliable, sector‑specific document pipelines.

  • aigpu
  • ai gpu
  • ai gpu cloud
  • aigpu dubai
  • document processing
  • enterprise ai
  • seed funding

By AiGpu Editorial · Editorial rewrite based on public reporting (TechCrunch AI)

← All articles