Luel
Category: AI Infrastructure
A rights-cleared multimodal AI training data marketplace and collection engine connecting a global contributor network with frontier AI teams. Luel was founded in 2025. The company is led by William Namgyal. Based in San Francisco, California, USA. Team size: 11-50. Total funding raised: $31.7M. Latest round: Seed. Key investors include General Catalyst, Lightspeed Venture Partners, Paul Graham, Orange Collective, Human Capital, Mo Amdani, SV Angel, Spot VC, Y Combinator.
- Founded
- 2025
- Headquarters
- San Francisco, California, USA
- Team size
- 11-50
- Total funding
- $31.7M
Value proposition
Luel is a two-sided marketplace for rights-cleared multimodal training data. AI teams submit dataset specs, Luel mobilizes a global contributor network of 600K+ people across 96 countries to collect it, runs multi-stage QA, and delivers audit-ready datasets with full provenance and consent records within days.
Products and solutions
Custom dataset collections (bespoke datasets built to spec with custom annotations), Off-the-shelf catalog (pre-collected datasets available for re-licensing), Third-party marketplace listings (external labs post datasets for revenue share). Datasets span egocentric robotics video, computer-use data, multilingual conversational speech, monologue speech, music libraries, and large imagery datasets.
Unique value
Speed (new campaigns spin up in under 24 hours, delivery within days vs. months from legacy vendors); Scale (600K+ contributors across 96 countries); Rights-cleared by design (consent evidence, chain-of-title, QA logs built in for enterprise compliance); Multi-modal breadth (video, audio, screen recordings, sensor streams, images, annotations); Proprietary QA pipeline; Sensor-aligned data for embodied AI (IMU, device pose, hand-object interaction).
Target customer
Frontier AI labs, generative AI companies, robotics companies, speech research teams, major social platforms, universities, hospitals, banks, and ML research groups needing high-quality, rights-cleared multimodal training data.
Industries served
AI/ML Training Data, Robotics, Voice/Speech AI, Computer Vision, Healthcare, Finance, Social Media, Academia
Technology advantage
Proprietary multi-stage QA pipeline; Global contributor network with behavioral incentive systems; Sensor-aligned data capture for embodied AI (physics layer with IMU data); Full provenance and consent records built into every dataset; Ability to spin up new dataset campaigns in under 24 hours; Business model compounds across three revenue streams (bespoke, catalog re-licensing, third-party marketplace).
How they differentiate
Unlike Kled (more consumer-app focused with crypto-powered incentives), Luel emphasizes enterprise workflow, speed (days vs. months), and a proprietary QA pipeline. Unlike Scale AI (focused on expert annotation/RLHF at ~$85/hr), Luel targets high-volume everyday multimodal data across dozens of languages at lower cost. Unlike Mercor (AI staffing marketplace), Luel is a data marketplace not a labor marketplace. Unlike Appen (legacy vendor), Luel offers modern, fast, multimodal collection with full provenance.
Main competitors
Kled, Scale AI, Mercor, Appen
Key partnerships
Y Combinator (W26 batch), General Catalyst (lead seed investor), Lightspeed Venture Partners (lead seed investor), Paul Graham (advisor/investor)
Notable customers
Generative AI labs (frontier model builders), Robotics companies, Major social platforms, Universities, Hospitals, Banks, ML research groups (specific names not publicly disclosed)
Major milestones
Founded late 2025 by William Namgyal (18) and Inigo Lenderking (19), Accepted into Y Combinator Winter 2026 batch, Launched publicly February 2026, Achieved ~$2M ARR within 6 weeks, Raised $31.2M seed round (May 2026) — one of largest seed rounds in YC history, Turned down 3 acquisition offers during raise, Processed 1M+ submissions through QA, Scaled to 600K+ contributors across 96 countries
Growth metrics
~$2M ARR within 6 weeks of founding; 600K+ contributors worldwide; 96+ countries covered; 1M+ submissions through QA pipeline; 40+ active dataset campaigns at any given time; 15M+ hours recorded
Market positioning
Positioned as the "global ground truth infrastructure for human data" — a neutral, two-sided marketplace bridging the gap between exhausted public web data and the need for rights-cleared, high-quality multimodal training data for frontier AI. One of the largest seed rounds in YC history ($31.2M), signaling strong investor conviction in the AI training data infrastructure layer.
Geographic focus
Global — contributors in 96 countries; HQ in San Francisco, USA; strong focus on under-served languages and geographies (South Asian patient-doctor conversations, Pakistani street Urdu, Japanese speech, etc.)
About William Namgyal
Ex-Founding Engineer at ezML (computer vision startup, previous exit); Ex-Founding Engineer & GTM Lead at Relixir (YC X25); LLM Security Research Intern at Northeastern University PEACH Lab; USACO Platinum-level competitive programmer at 16; Berkeley M.E.T. Dropout
Official website: https://www.luel.ai/