Uber integrates OpenAI frontier models into app to assist drivers and speed bookings

Uber is integrating OpenAI frontier models into its app to provide in‑app AI assistants and new voice features aimed at helping drivers increase earnings and enabling riders to book trips more quickly. The company says the technology is designed to reason across real‑time variables in its marketplace; the change matters because those variables directly affect trip availability, pricing and routing decisions for millions of users and drivers. According to Uber, the models will power conversational assistants inside the app and voice interactions that can surface timely guidance to drivers and faster booking flows for riders. Those in‑app assistants are intended to present information and suggestions in context, so drivers and riders get relevant options without switching screens or digging through menus.

Uber says the models are built to reason across a range of operational signals: traffic conditions, weather, airport arrival schedules, local events and shifting demand across the platform. By considering these factors together rather than in isolation, the company expects the system to generate recommendations that reflect the current state of supply and demand in a given area. The integration is being applied across Uber’s global marketplace, which the company describes as handling roughly 40 million trips per day and encompassing about 10 million drivers and couriers. Those participants operate across some 15,000 cities in more than 70 countries, making the deployment a scale test of the models’ ability to work in many different markets and conditions.

For drivers, Uber frames the update as a tool to help them make smarter decisions about where and when to position themselves, which trips to accept and how to navigate changing conditions. For riders, the voice features and assistant workflows aim to reduce friction in booking and provide faster responses when demand spikes or travel conditions change. Uber emphasizes that the models will reason in real time against marketplace signals rather than operate as static rules. That approach is intended to let the system adapt to sudden changes — like unexpected weather or a surge of airport arrivals — so the guidance drivers receive and the options presented to riders reflect current conditions.