Uber Leverages OpenAI to Simplify Earning for 10 Million Drivers

Uber is now using OpenAI’s models to operate a marketplace at a large scale, processing 40 million trips daily across 15,000 cities and 70 countries. The collaboration is delivering AI-powered guidance directly to the company’s 10 million drivers and couriers, answering questions about optimal positioning and earnings potential; drivers can now ask questions in plain language and receive tailored responses. This shift is enabled by a new AI assistant designed to simplify earning opportunities and reduce friction for riders, allowing Uber to rapidly deploy streamlined experiences. “For the first time, technology is leading what can be solved. Problems that once felt out of reach are now possible to address,” says Aarathi Vidyasagar, VP of Engineering and Science at Uber, signaling a fundamental change in the company’s technological capabilities.

Uber Assistant Provides Real-Time Driver Earnings Guidance

Uber operates at a large scale: 40 million trips per day, with 10 million drivers and couriers across 15,000 cities in over 70 countries. Beyond matching riders and drivers, the company is leveraging artificial intelligence to actively manage the complexities of real-time conditions, including traffic patterns, weather events, and localized demand fluctuations across 15,000 cities and 70 countries. This isn’t a limited pilot program; it’s the engine powering a massive daily operation, allowing Uber to analyze previously complex data sets. A key result of this integration is the Uber Assistant, an AI-powered tool designed to provide drivers with actionable guidance.

Drivers, who operate with varying schedules and priorities, can now receive answers to questions like “Where should I position myself right now?” and “Is the airport worth driving to?” According to Uber’s Director of Product Management, Dharmin Parikh, the Assistant aims to “enable drivers to make better decisions for themselves by providing a summarized view of the marketplace and real-time insights.” The system translates complex data, such as earnings trends and heatmaps, into easily understandable positioning suggestions, allowing drivers to ask follow-up questions and refine their strategies. Uber discovered that this assistance significantly accelerates the learning curve for new drivers, helping them grasp platform dynamics far more quickly than through trial and error. This level of proactive support represents a significant leap in Uber’s technological capabilities.

This includes utilizing faster models for simple tasks and larger reasoning models for more complex queries, all governed by an internal “AI Guard” to ensure safety, privacy, and policy compliance. Parikh explains that the goal is to build a trustworthy system, stating, “If users don’t trust the system, you lose them quickly. But when they see value, they return.” Early results from the U.S. driver network show drivers are repeatedly returning to ask follow-up questions and optimize their time on the platform, validating the product as a long-term utility, not just an onboarding tool, and demonstrating improved time utilization and stronger engagement with the platform.

Multi-Agent AI System Prioritizes Safety and Low Latency

Uber’s integration of OpenAI models extends beyond managing the complexities of its global marketplace; a core focus has been building an AI system prioritizing accuracy, safety, and speed. This led to the development of a multi-agent system, designed to route user requests to specialized systems optimized for specific tasks, ensuring efficient and appropriate responses. For instance, inquiries regarding earnings are handled separately from onboarding questions, leveraging different models suited to each need. The system employs a tiered approach, utilizing faster, smaller models for straightforward classifications and rapid responses, while reserving larger, more powerful reasoning models for complex tasks. Crucially, Uber implemented “AI Guard,” an internal governance layer designed to screen prompts and responses, promoting safety, privacy, and security, and minimizing inaccuracies. The company found that providing accurate, useful recommendations encourages repeat engagement and increased productive time on the platform.

OpenAI Realtime APIs Enable Hands-Free Voice Booking

Uber is rapidly integrating OpenAI’s Realtime APIs to fundamentally alter how its marketplace operates, moving beyond simple machine learning applications to leverage the reasoning capabilities of large language models. “The Assistant is helping drivers ramp up quickly, compared to taking several hundred trips to understand how the platform works,” says Parikh. Beyond driver assistance, Uber is also pioneering voice-activated booking through the integration of OpenAI Realtime APIs. This move addresses the limitations of traditional app interfaces, particularly for complex requests or users with accessibility needs.

Riders can now articulate their needs in natural language, for example, specifying luggage requirements and passenger numbers, and receive tailored recommendations. “Voice removes the barrier of completing one task at a time,” says Parikh. “You can express full intent naturally, and the system can orchestrate the outcome.” This hands-free functionality extends to drivers as well, allowing for safer and more efficient interaction with the app while on the road. Vidyasagar emphasizes the broader implications, stating, “Voice removes the multi-tap barrier because you can say multiple things. It unlocks that ability to connect the various parts of the ecosystem.”

LLMs Foster Collaboration & Accelerated Product Iteration

This isn’t simply about automating existing processes; the company is experiencing a fundamental shift in its technological capabilities, allowing it to tackle previously complex problems. A key example is the development of Uber Assistant, an AI-powered tool initially designed to accelerate the learning curve for new drivers. The system transforms complex marketplace data, earnings trends, heatmaps, and real-time demand, into actionable insights, helping drivers optimize their positioning and earning potential. While anticipated to benefit newer drivers most, experienced drivers are also repeatedly utilizing the Assistant, demonstrating its value as a long-term utility. Uber’s approach to implementing this AI system prioritizes accuracy, safety, and low latency, achieved through a multi-agent architecture. An internal governance layer, termed AI Guard, further ensures responses adhere to policy, reduce inaccuracies, and maintain consistency.