Uber is giving ChatGPT Work to its entire global workforce of about 30,000 employees, expanding a pilot that began with 1,000 staff earlier this year into a company-wide deployment. Uber and OpenAI announced the expansion on Thursday, October 22, describing it as Uber’s first company-wide rollout of an agentic platform built for knowledge work, running inside ChatGPT Enterprise.
The move takes OpenAI’s enterprise agent well beyond a trial. The pilot drew staff from across finance, operations and product, and about 80 percent of those early users were not engineers, Uber’s vice president of engineering Sumanth Sukumar told TechCrunch. The company-wide rollout adds marketing and the rest of the business, supported by an internal AI skills library, a set of approved integrations and dedicated employee training.
ChatGPT Work at Uber: From a 1,000-Person Pilot to 30,000 Seats
ChatGPT Work is the agent OpenAI introduced in July, built to take on complex, multi-step tasks and deliver finished work rather than a chat answer. At Uber it connects to the tools staff already use, including Google Drive, Gmail, Google Calendar and Slack. OpenAI says those connections let employees research information, prepare meeting briefs, create presentations, analyse data and complete workflows that cross several applications.
Sukumar said pilot users settled into three broad patterns. Teams analysed complex operations data to identify trends and develop recommendations, localised training materials for different markets and languages, and built personalised interactive dashboards to manage their daily work. Uber has not prescribed a single way to use the agent now that it is everywhere. “Our approach is to make powerful AI tools available across Uber while giving teams room to experiment,” Sukumar said.
One earlier experiment points to how Uber expects usage to spread. This year the company paired 30 AI engineers with teams across the business and gave each pairing two weeks to automate key parts of the team’s job. Sukumar said those “AI pods” showed how much opportunity appears when technical expertise is combined with a deep understanding of a particular role.
GPT-6.1 Sol as the Default, Codex for the Engineers
The deployment runs on OpenAI’s current model line. GPT-6.1 Sol is the default model inside ChatGPT Work for Uber staff, a model OpenAI positions as balancing frontier performance with cost efficiency. Employees also get OpenAI’s multimodal tools, including voice and image generation, so teams can pick the model and tool that fit the task. Uber’s developers and software engineers will keep using Codex for software development and coding work rather than the general agent.
Uber’s chief technology officer Praveen Neppalli Naga framed the rollout around capability rather than a fixed use case. “As AI transforms how we work, we want to give everyone the tools to innovate, solve problems and help shape Uber’s future,” he said in OpenAI’s announcement. OpenAI president Greg Brockman called the expansion “a big step toward expanding access to frontier AI capabilities and tools.”
Uber did not disclose what the company-wide rollout costs. Sukumar said the company will watch how employees use the technology, what workflows they develop and whether it helps them work more effectively, and will use those findings to guide how it deploys AI across the business. The two companies say they plan to share more on adoption, use cases and measurable outcomes as the rollout progresses. The enterprise-wide pattern is becoming familiar: Microsoft Copilot recently reached more than 330,000 Volkswagen Group employees in a similarly broad deployment.
Why It Matters
ChatGPT Work’s value to OpenAI is measured less in any single feature than in seats like these. A 30,000-person rollout at a company of Uber’s scale is the kind of deployment that turns an agent from a product announcement into workplace infrastructure, and it lands in the middle of a direct contest with Anthropic, whose Claude Duet proactive agent reached research preview this week aimed at the same professionals. The detail that four in five pilot users were not engineers matters for both companies: the market for workplace agents is the whole staff, not the engineering department, and Uber is now a reference customer for that claim. For other large employers, the open-ended design is the test to watch — Uber is deliberately not mandating workflows, so the usage data it collects will show whether general agents find their own work inside a big company or stall without direction. Retail is running a parallel experiment in public, where Walmart has built agent profiles and a gateway for AI shopping agents rather than blocking them.
Background
Uber and OpenAI have worked together well before this rollout. OpenAI’s earlier customer story on the partnership described Uber using its models for driver- and rider-facing products, including Uber Assistant, which turns marketplace data such as earnings trends and heatmaps into guidance for drivers, and voice booking built on the Realtime API. ChatGPT Work itself launched on July 9 as an agent inside ChatGPT that can act across a user’s apps and files and stay with a project for hours, initially for Pro, Enterprise and Edu plans. Thursday’s expansion moves the relationship from customer-facing products to Uber’s own internal workforce.
Source: TechCrunch, Unite.AI reporting OpenAI’s announcement
0 Comments