Uber is expanding its AI and data services unit, Uber AI Solutions, as it looks to support labs and enterprises looking to build AI models and deploy agents.
The company is offering the data and AI platform it uses internally to enterprises in a move they rhymes with what Amazon and Google do with cloud computing. Build the expertise and platform for internal use, then turn it into a business.
As previously noted, Uber is more of a data company than one focused on mobility. Uber's expertise is in collecting, labeling, testing and localizing data for its operations and then optimizing interfaces to add value. As generative and agentic AI take hold, these data services matter a lot more.
- Uber aims to press its 'data advantage' for AI, model training
- How Uber's tech stack, datasets drive AI, experience, growth
- Uber Q4 strong, eyes platform expansion
Uber's core pitch for its platform: "As we’ve scaled Uber to power more than 33 million trips across mobility and delivery every day, we have invested in innovation in product, platform, and artificial intelligence (AI) and machine learning (ML) . To enable these, we’ve created a world-class technology platform that is designed to meet our evolving requirements across data labeling, testing, and localization. We’re now making this available."
Here's what Uber AI Solutions is rolling out:
- Global digital task platform, which connects enterprises to experts in coding, finance, law, science and linguistics. Tasks include annotation, translation and editing for multi-modal content. Think Uber gigs expanded broadly.
- Uber data foundry, a service that provides packaged and custom datasets including audio, video, image and text to train large language models (LLMs).
- Infrastructure for AI. Uber said it is making its platforms to manage data annotation projects and validate AI outputs available to enterprises.
- An interface designed to "become the human intelligence layer for AI development worldwide." According to Uber, the interface will allow enterprises to describe data needs in plain language for setup, tasks, workflow optimization and quality management.
