What Luel does

Luel Inc. operates a two-sided AI training data marketplace connecting a global supply base of 500,000+ contributors across 96 countries with enterprise AI developers and frontier AI labs. The platform delivers multi-modal training data spanning audio, video, image, and sensor inputs, with a current catalog of 30 production-ready datasets covering egocentric video (20,000+ unique tasks), multilingual conversational speech (English, French, German, Japanese), doctor-patient consultation (1,000+ hours under HIPAA), Telugu TTS (900+ hours), monologue speech, music, and computer-use workflows. Customers include 30+ frontier AI labs and enterprise AI procurement teams, and the company has reported approximately $2M ARR within six weeks of its YC W26 Demo Day.

The platform is built on a proprietary continuous data engine with a three-part Luel Certified review covering rights clearance, provenance, and quality. Technical components include biometric voiceprint-based contributor session verification, LiveKit integration for stereo speaker separation, and synchronized IMU plus egocentric video capture for robotics and physical-AI applications. The company has formalized compliance with HIPAA, GDPR, CCPA, multiple US state privacy laws, and biometric privacy statutes (BIPA, CUBI, Washington) via a published Biometric Policy. The business operates on three revenue streams: a 15% marketplace commission (with contributors receiving approximately $0.50 per minute for video/voice), quote-based bespoke dataset collection, and enterprise data licensing on multi-year contracts. Luel is headquartered in San Francisco, was founded by William Namgyal (CEO) and Inigo Lenderking (COO), launched as part of the Y Combinator W26 batch in February 2026, and raised a $31.2M seed round in May 2026 led by Lightspeed, General Catalyst, and YC.

Name

Luel

Legal name

Luel Inc.

Website

https://luel.ai

Company type

Private

Founded year

2026

Operating status

Operating

Headcount range

1–10 employees

Short description

Luel Inc. operates a two-sided AI training data marketplace connecting 500,000+ contributors across 96 countries with frontier AI labs and enterprise buyers, supplying multi-modal datasets spanning audio, video, image, and sensor inputs.

Ownership category

akta.pro rank

Product category

AI Training Data Marketplace

NAICS

Software Publishers (513210), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182)

SIC

Services-Prepackaged Software (7372)

akta.pro primary industry

AI Integration & Orchestration Platforms (Connectors, Workflow, iPaaS for AI) (HDAEANAI)

akta.pro secondary industry

AI Tutoring Systems (Conversational Tutors, Homework Helpers) (EDAFAFAC)

Keywords

• AI training data
• Multimodal datasets
• Data marketplace
• Rights-cleared data
• AI data licensing

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices1 record

Markets served

GTM type

B2B and B2C

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Infrastructure, Supply Chain

Revenue model

1. Marketplace dataset licensing commission: Luel retains 15% of net proceeds from sublicensing supplier-listed datasets to enterprise buyers; suppliers receive 85%. Third-party payment processing fees are deducted before the split. Payouts to suppliers typically occur within 30 days of Luel receiving buyer funds.
2. Bespoke dataset collection fees: Enterprise buyers submit dataset specifications (modality, scenario, devices, QA rules); Luel mobilizes vetted contributors, runs multi-stage QA, and delivers a structured licensed dataset within days. Quote-based pricing per custom engagement.
3. Data licensing (Enterprise sales of User Content): Enterprises purchase/license User Content through the Service for AI training, fine-tuning, evaluation, and commercial deployment of derivative models. Sales are final and irrevocable; all fees paid in advance, non-cancelable and non-refundable except as required by law.
4. Contributor payouts (cost of revenue, not income): Luel pays contributors per submission based on duration and listing payout rates; payouts are determined only when a submission is marked payable, with fraud/chargeback clawback rights retained.

Pricing tiers

Go-to-market motion1 record

Distribution channels4 records

Marketing channels6 records

Core offering

Luel operates a two-sided multimodal AI training data marketplace that connects individual Contributors with AI Enterprises. The platform supplies rights-cleared, provenance-tracked training datasets spanning audio, video, image, and IMU/sensor modalities, delivered either as off-the-shelf Luel Certified catalog items or as bespoke collections produced by a vetted global contributor network of 500,000+ contributors across 96 countries. Built on a proprietary continuous data engine with multi-stage QA, voiceprint identity verification, and LiveKit stereo speaker separation, the platform serves frontier AI labs building foundation models, voice agents, robotics policies, and embodied agents.

Product overview

Luel is a single unified product offered as a two-sided marketplace platform with companion modules. The core product is the Luel AI Training Data Marketplace — a multi-modal lab that connects individual Contributors with AI Enterprises, built on a proprietary continuous data engine. The platform pairs an off-the-shelf Data Marketplace catalog (30 Luel Certified datasets including Computer-Use Workflows, General Egocentric Video, Music Library, eight speech/conversational/monologue datasets across English/French/German/Japanese, Telugu Expressive TTS Voice, and Doctor-Patient Consultation) with a global Collection Engine (500,000+ contributors across 96 countries) that runs bespoke custom collections on the same rails. Modules on the platform include the Supplier Portal for listing datasets (Suppliers retain 85% of net proceeds), the Bespoke/Custom Collection Service for scoped-to-spec datasets, and companion iOS and Android Mobile Apps for Contributors. Each dataset is gated by the Luel Certified three-part review (rights-cleared, provenance-tracked, quality-audited) and ships with consent evidence, chain-of-title, and audit logs.

Differentiator

Problem solved

Functional benefit

Products and services

• Data Marketplace Production-ready dataset catalog where enterprises can search, license, and analyze 30 datasets spanning audio, sensor, and video categories. Each listing ships with consent evidence, provenance, chain-of-title, and audit logs under the Luel Certified three-part review.
• Collection Engine Vetted global contributor network of 500,000+ contributors across 96 countries that produces datasets to a buyer's specification, scoped end-to-end with QA, recruitment, payouts, and delivery. The same pipeline that licenses off-the-shelf datasets also runs custom collections, typically delivering bespoke datasets within days.
• Supplier Portal Portal for approved suppliers (individuals and companies) to list datasets on the Luel Marketplace. Suppliers grant Luel a non-exclusive worldwide license to access, QA, sample, market, negotiate, and sublicense datasets to enterprise buyers while retaining ownership of underlying data. Suppliers receive 85% of net proceeds and Luel retains 15%.
• Luel Mobile Apps (iOS and Android) Companion mobile applications distributed through the Apple App Store and Google Play Store, providing contributors with on-device access to the marketplace for submitting video, audio, voice recordings, images, documents, and other content.
• General Egocentric Video dataset First-person, head-mounted-style video recordings spanning 20,000+ unique tasks across households, factories, shops, and many more environments. Every clip is paired with synchronised IMU data (accelerometer + gyroscope), approximately 95% hand visibility, and clear field-of-view. Annotated with scene/location/lighting/motion metadata and optional activity/sub-step labels; designed for embodied AI and robotics training.
• Computer-Use Workflows dataset Continuous screen recordings of real practitioners working in 500+ professional desktop and web applications including CAD/CAE, BI/analytics, design, video/audio, ERP/CRM, healthcare, and bespoke enterprise dashboards. Capture includes cursor traces, software/version detection, OS metadata, and per-clip workflow boundaries for training computer-use agents.
• English Conversational Speech dataset Stereo multi-speaker English dialogue recordings with L/R speaker separation (recorded via LiveKit with isolated per-participant tracks). Includes casual dialogues, structured multi-participant meetings, and professional interactions across diverse topics with speaker diarization, word-level timestamps, and per-utterance emotion labels across 18 categories.
• Telugu Expressive TTS Voice dataset Comprehensive TTS training dataset of natural Telugu speech recordings (900+ hours) from native speakers across major Telugu-speaking regions. Includes phoneme-level alignment, emotion and style coverage, phonetically balanced scripts, and regional dialect variants for training natural-sounding Telugu text-to-speech models.
• Doctor-Patient Consultation dataset Clinical consultation dialogues (1,000+ hours) between doctors and patients across diverse hospital settings (surgeons, endocrinologists, cardiologists, neurologists, etc.). Covers initial visits, follow-ups, diagnosis discussions, and treatment planning with structured medical metadata; WAV audio (44.1kHz to 48kHz), JSON transcripts with medical terminology, and CSV metadata by department classification. Marked as HIPAA-compliant data handling.
• Music Library dataset Large-scale, professionally produced music collection spanning multiple modern genres (vocal-focused tracks and songkits, house, trap, afrobeat, indie/shoegaze) with full song construction kits (separated stems, looped phrases, one-shot hits, per-kit MIDI) and production toolkits. Distributed under gated access for evaluation, internal research, and partner integrations.
• Bespoke / Custom Collection Service Service for enterprises to request a dataset spec (modality, scenario, instructions, devices, QA rules). Luel posts the listing, matches vetted contributors, runs submissions through multi-stage QA, and delivers a structured licensed dataset, typically within days, with manifest, sensor channels, annotations, provenance, and consent documentation.

Quantifiable outcome

• ~$2 million ARR within six weeks of YC W26 Demo Day
• +5 more outcomes

Segments5 records

Ideal customer profiles5 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration6 records

AI capability8 records

Feature7 records

Partnerships

Eight partnerships are on record, tiered minor and core.

• Kled (Nitrility Inc.)minorOthers · 21 May 2026Competitive-adjacent relationship in the human-generated AI training data marketplace category; Kled founder Avi Patel publicly accused Luel of copying Kled's marketplace idea after Luel's $31M seed announcement, sparking an attention-economy controversy. Listed by third-party reviewers as one of six alternatives to Luel AI.
• Waffle VideominorOthersPeer platform in the same human-generated robot-training-data gig category; profiled alongside Luel and Kled in a Wired feature on the emerging 'gig economy' of robot training data.
• StripecoreTechnology or IntegrationThird-party payment processor used by Luel to process contributor payouts and enterprise purchases; payment data is governed by Stripe's privacy notice.
• LiveKitcoreTechnology or IntegrationConversational speech datasets (English, German, etc.) are recorded via LiveKit with isolated per-participant audio tracks delivered as stereo L/R-channel WAV files for downstream speaker separation training.
• LabellerrminorOthersListed alongside Luel as one of seven top data labeling companies in robotics for 2026; indirect industry positioning comparison.
• JAMSminorOthersDesignated arbitration provider in Luel's Terms of Service for resolving disputes on an individual basis (mandatory individual arbitration).
• Apple, Inc.coreOEM/ Whitelabel/ Licensing PartneriOS App distributed via Apple App Store under Apple's then-current Apple Media Services Terms; Apple is a third-party beneficiary of the Terms of Service for iOS app licensing.
• Google LLCcoreOEM/ Whitelabel/ Licensing PartnerAndroid App distributed via Google Play Store under Google Play Terms of Service; Google is a third-party beneficiary of the Terms of Service for Android app licensing.

Scale indicators11 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Labelbox: Labelbox provides a data labeling and annotation platform with a contributor marketplace used by AI teams across foundation models, robotics, and autonomous systems. It is a direct competitor to Luel for multimodal annotation and dataset curation workflows at enterprise AI buyers.
• Sama: Sama provides ethically sourced data annotation, data collection, and model validation services for AI applications including computer vision and speech. It competes with Luel on enterprise multimodal annotation contracts and is recognized for its worker-welfare positioning adjacent to Luel's rights-cleared stance.
• Scale AI: Scale AI operates a large, enterprise-focused data labeling and annotation platform serving many of the same frontier AI labs (and adjacent robotics/autonomous-vehicle customers). Luel competes with Scale directly for multimodal, rights-cleared training-data contracts and the same enterprise procurement relationships.
• Toloka: Toloka (a Nebius company) runs a large, global crowdsourced data labeling and collection platform serving foundation-model developers. It is a direct peer in the human-generated, rights-cleared training-data marketplace category where Luel is competing.
• Surge AI: Surge AI runs a contributor-driven data labeling and evaluation marketplace serving frontier foundation-model developers. It overlaps with Luel's supply-side model (vetted global contributor network) and the same frontier-AI buyer cohort for high-quality human-generated training and evaluation data.
• Defined.ai: Defined.ai operates an AI training-data marketplace with rights-cleared speech, text, image, and video datasets for enterprise AI buyers. It is a direct peer that competes for similar procurement-grade dataset contracts and emphasizes ethical sourcing — a positioning similar to Luel's 'Luel Certified' framing.
• Waffle: Waffle is an emerging peer in the human-generated robot-training-data gig category, profiled alongside Luel and Kled in a Wired feature. It targets the same embodied-AI / robotics buyer cohort with contributor-supplied video and sensor data.
• Kled: Kled is a closely adjacent human-generated AI training-data marketplace; its founder publicly accused Luel of copying its marketplace model, and the two are routinely profiled together as competing platforms in the same AI-training gig category.

Broad incumbents

• Appen: Appen is a publicly listed incumbent that supplies annotated training data (speech, text, image, video) to large enterprise and AI customers. Although broader and more general-purpose than Luel, it competes for the same enterprise speech and multimodal dataset budgets and has the largest incumbent footprint.
• Telus International (TELUS AI): Telus International is a large, broad-based digital CX and AI data services provider offering annotation, data collection, and evaluation across modalities. While far larger and more diversified, it competes with Luel for enterprise AI dataset RFPs.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/luel

Compliance7 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles2 records

Funding overview

Funding rounds2 records

Investors8 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Luel