NuMind is a Paris-based, Y Combinator-backed AI company building multimodal large language models that turn messy documents - PDFs, scans, spreadsheets, handwritten forms - into clean, structured JSON. Its NuExtract platform and open-source models let developers and enterprises extract exactly the fields they define, in any language, either as a hosted API or deployed privately on their own infrastructure.
OnDeck AI is a Vancouver-based Y Combinator (S25) company building Perception-0, a vision model that lets organizations find any object, behavior, or event in video footage using plain language - without collecting training data or building a custom computer-vision pipeline. Started as a fisheries-monitoring project, it now serves defense agencies, robotics teams, and security operators, including a pilot with Singapore's Ministry of Defence.
Unisson (YC W26) is a San Francisco startup building AI agents that act as hands-on product experts for customer-facing teams at B2B software companies. Its agents learn to use a client's product in about 15 minutes by exploring the interface and codebase, then work alongside implementation, customer success, and sales engineering teams inside Slack, email, and text to run onboarding, migrations, integrations, audits, and change management. Founded in 2026 by Varun Mathur and Tom Achache, the company charges only when its agents are actively doing work.
Reducto is a San Francisco AI company that turns messy, complex documents into clean, LLM-ready data. Its agentic document platform pairs traditional computer vision with vision-language models and a multi-pass 'Agentic OCR' correction loop to parse, split, extract, classify, and even edit PDFs and other formats through a developer API. Founded in 2023 by MIT grads Adit Abraham and Raunak Chowdhuri, Reducto has processed billions of pages for AI teams and enterprises in finance, healthcare, legal, insurance, and compliance, and raised $108M total including a $75M Series B led by Andreessen Horowitz in October 2025.
Oumi (Open Universal Machine Intelligence) is a Kirkland/Bellevue, Washington-based public benefit corporation building a fully open-source, end-to-end platform for foundation models. Founded in 2024 by ex-Google, Apple, Meta and Microsoft engineers, its open software stack lets researchers and enterprises prepare data, pretrain, fine-tune, evaluate and deploy LLMs and vision-language models - from 10M to 405B parameters - anywhere from a laptop to a cloud cluster. Backed by $10M in seed funding and a coalition of 13 top research universities, Oumi pitches itself as the 'Linux of AI': a neutral, collaborative, unconditionally open alternative to closed frontier labs.
Pulse is a San Francisco AI company building production-grade document extraction for enterprises. Its vision-language models and OCR pipeline turn messy PDFs, scans, and spreadsheets into clean, structured, LLM-ready data - powering RAG, automation, and analytics across finance, healthcare, insurance, legal, and supply chain.
Thoth AI is a Singapore-headquartered global AI data solutions company that builds the high-quality datasets, evaluation frameworks, and human-in-the-loop operations that production AI systems depend on. Spanning 200+ languages and 100+ countries, it handles data collection and annotation across images, video, 3D point cloud, text, and audio, along with RLHF, model evaluation, content moderation, and multilingual customer experience for teams building LLMs, VLMs, multimodal systems, and applied AI like robotics.