ai-training-data
(16)
micro1 Bet the AI Boom Would Run Out of Human Judgment
The young company began with an AI interviewer. It is now selling the scarce ingredient frontier models, workplace agents and robots all need: structured evidence of how capable people think and work.
Shofo
Shofo is a San Francisco startup (Y Combinator W26) building what it calls the world's largest indexed video library - a 'Common Crawl for video.' It continuously crawls billions of public short-form clips from platforms like TikTok, Instagram Reels and YouTube Shorts, then runs them through a pipeline that sanitizes, deduplicates, segments and labels footage with object detection, activity recognition and reasoning annotations. Shofo sells custom, ready-to-train datasets to AI labs, delivering a request like '100K hours of cooking videos where someone is holding a pan' in days rather than months.
Integral Privacy Technologies
Integral is a privacy engineering company that turns sensitive real-world data - starting with HIPAA-regulated healthcare records - into compliant, AI-ready datasets. Founded in 2022 by Shubh Sinha and John Kuhn, it pairs automated de-identification and continuous risk assessment with embedded statisticians and privacy engineers (its Forward Deployed Privacy Services) so enterprises and AI labs can use regulated data quickly without violating privacy law. The company positions itself as an independent privacy layer for the real-world data economy and has raised roughly $24.9M to date, including an $18M Series A closed in July 2026.
Protege
Protege is a New York-based AI data platform that connects owners of private, real-world data - hospitals, media libraries, motion-capture studios and more - with the AI labs and enterprises that need it to train, fine-tune and evaluate models. It licenses and curates proprietary datasets under strict privacy, IP and compliance controls, so data holders get paid and provenance is preserved. Founded in 2024 by Bobby Samuels and Travis May, Protege has assembled access to billions of clinical notes, hundreds of millions of medical images, and hundreds of thousands of hours of video and audio, positioning itself as core infrastructure for the post-web era of AI development.
iMerit Technology
iMerit is a global AI data solutions company that supplies the labeled, enriched and expert-refined data used to train, fine-tune and evaluate machine learning and generative AI models. Founded in 2012 by Radha Basu, it pairs a specialized human workforce with its Ango Hub annotation platform and its Scholars network of domain experts to serve Fortune 500 clients across autonomous vehicles, healthcare, geospatial and financial services. In June 2026, enterprise-AI firm EXL agreed to acquire iMerit for up to $310 million.
Stardust AI
Stardust AI is a San Francisco-founded, data-centric AI company that helps enterprises turn raw, messy, multimodal data into the high-quality datasets that machine-learning models actually need. Through its Rosetta auto-labeling platform, its MorningStar data engine and its COSMO dataset offerings, Stardust annotates and manages 2D/3D/4D data spanning images, video, point clouds, audio and text - with heavy expertise in autonomous driving and large language models. The company works with names like SAIC Motor, BYD, Geely, Bosch, ZF, Xiaomi and JD.com, and pitches a simple thesis for the AI era: whoever controls the data controls the model.
Created by Humans
Created by Humans is a San Francisco startup building a licensing marketplace that lets authors and other creators sell the rights to their work for AI training, reference, and transformation - and get paid for it. Founded by former Scribd CEO Trip Adler, the platform lets rights holders claim their books (by ISBN or upload), set the terms under which AI companies may use them, and track usage and payments, positioning licensing as an alternative to lawsuits between creators and AI developers.
Pareto.AI
Pareto.AI is a San Francisco-based, talent-first human data platform that connects AI labs with a deeply vetted network of expert labelers to produce premium training data, RLHF, and evaluation signals. Founded by Thiel Fellow Phoebe Yao, the company started as a bootcamp for women and work-from-home mothers, pivoted into AI data labeling in 2023, and now positions itself as the 'verification layer' for reinforcement learning on real-world expertise, turning specialist judgment into durable reward signals under the banner 'Expert data for AGI.'
Bifrost AI
Bifrost AI builds simulation and evaluation infrastructure for physical AI. Its platform generates photorealistic 3D synthetic data and stress-tests robotics and computer-vision systems across thousands of simulated scenarios - weather, sensor noise, rare edge cases - so teams can train and validate models in minutes instead of months. Used across maritime, geospatial, aerial, off-road, industrial, and off-world domains, with collaborators including NASA JPL and the U.S. Air Force. Founded in 2020 and backed by an $8M Series A led by Carbide Ventures with participation from Airbus Ventures and Peak XV's Surge.
Thoth AI
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.
Wow AI
Wow AI is a US-registered, end-to-end AI training-data company that supplies high-quality, multilingual datasets and human-in-the-loop services - data collection, transcription, annotation, validation and RLHF - to teams building large language models, voice assistants and computer-vision systems. Its crowdsourcing network spans 170,000+ contributors across 120+ languages, with 100,000+ hours of audio and domain datasets for finance, healthcare and retail. The founding team has since extended the vision into AIxBlock, a decentralized platform for building, training and deploying AI, and WowDAO, a community-owned AI ecosystem.
LanceDB
LanceDB builds the data backbone for multimodal AI. Its open-source Lance columnar format and lakehouse let teams store, search, and train on text, images, video and embeddings in one system - replacing the brittle stack of Parquet files, vector stores and feature pipelines that AI teams usually stitch together. Used by Midjourney, Runway, Character.AI, WeRide and others, LanceDB raised a $30M Series A in 2025, bringing total funding to roughly $41M.
Centaur AI
Centaur AI (Centaur Labs) is a Boston-based healthcare data annotation platform that taps a network of 50,000+ vetted medical and scientific experts to label, evaluate, and monitor the multimodal data that powers medical AI. Born out of MIT's Center for Collective Intelligence, the company uses performance-weighted 'wisdom of the crowd' methods - delivered through its gamified DiagnosUs app - to produce labels that match or beat individual specialists, on a HIPAA-compliant, SOC 2 platform.
Mercor
Mercor is an AI-powered talent marketplace that recruits domain experts - doctors, lawyers, engineers, scientists - to train and evaluate frontier AI models for labs like OpenAI, Anthropic, Google, Meta, and Amazon. Founded in 2023 by three Thiel Fellows, it has grown from a hiring tool into the human data backbone of the AI economy.
Wirestock
Wirestock is a two-sided marketplace that connects 700,000+ photographers, videographers, illustrators, and 3D artists with AI labs that need ethically-sourced, high-quality multimodal training data. After pivoting from stock-content distribution in 2023, the company now supplies six of the largest foundation-model makers and is running at a $40M revenue run rate.
Surge AI
Surge AI is a San Francisco data annotation company that produces high-quality human-labeled datasets and RLHF feedback for the world's leading AI labs - OpenAI, Google, Anthropic, Meta and Microsoft among them. Founded in 2020 by ex-Twitter and ex-Facebook ML engineer Edwin Chen, it bootstrapped to roughly $1.2B in annual revenue with around 110 employees and a labeler network reported near one million.