Defined.ai Expands AI Training Data Marketplace
The platform now offers ten petabytes of licensed, consented data for specialized AI models.
1Marketplace Expansion Strategy
Defined.ai has launched an expanded marketplace for artificial intelligence training data, featuring nearly two thousand datasets totaling ten petabytes. Announced by CEO Daniela Braga at Web Summit, the platform aims to provide developers with licensed, anonymized, and consented information. This segment now accounts for 70% of the company's total revenue.
2AI-Powered Data Discovery
The platform integrates a large language model assistant to streamline how clients locate specific data. Users can describe their requirements using natural language in any language to receive tailored suggestions. This tool simplifies navigation through complex multimodal formats, including audio, video, text, and specialized industry-specific datasets for finance and healthcare.
3Strategic Access Programs
To lower barriers for developers, the company introduced the Strategic Access Program, offering research and commercial licensing tiers. The research license costs 20% of the commercial price, allowing firms to test data efficacy before committing to full purchases. This model addresses client uncertainty regarding how specific datasets improve model performance.
4Partner Acceleration Initiative
The Partner Acceleration Program invites content creators and companies to monetize their proprietary data through the platform. Defined.ai manages the licensing, promotion, and commercialization processes for these partners. All new contributors undergo rigorous identification, legal evaluation, and cybersecurity checks to ensure the integrity and quality of the marketplace offerings.
5Ethical Data Sourcing
Defined.ai strictly avoids web scraping, emphasizing that all provided information is obtained with explicit consent. The company pays data owners for their contributions, ensuring a transparent and ethical supply chain. By focusing on private, human-generated data, the firm differentiates itself from competitors relying on publicly available, non-consensual internet content.
6Shift Toward Specialized Data
Market demand is moving away from generic content toward highly specialized information for specific sectors. Daniela Braga notes significant interest in healthcare, finance, and physical AI. Models are evolving beyond general reasoning, requiring granular inputs like patient pathways, laboratory imaging, and complex physical task recordings to function effectively in real-world environments.
7Physical AI Requirements
Physical AI development now requires data beyond simple camera footage, incorporating sensor-based recordings from human movements. Systems are being trained on tasks like furniture assembly, automotive repair, and electrical work. By utilizing sensors on hands and chests, developers can capture the precise physical interactions necessary to train robots for real-world operations.
8Early Market Adoption
Initial metrics indicate strong interest, with 280 companies consulting the catalog within two weeks of the launch. The platform recorded 344 new registrations and 333 sample downloads during this period. These figures suggest that the industry is actively seeking high-quality, specialized datasets to refine their next generation of artificial intelligence models.
Original article: ECO — Defined.ai Expands AI Training Data Marketplace