AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Uber puts all 30,000 employees on ChatGPT Work

Collected Oct 22, 2026

Uber and OpenAI said Thursday that they are expanding their partnership, and Uber told TechCrunch that its whole global workforce — roughly 30,000 employees — will now have access to ChatGPT Work. The move follows a smaller trial run earlier this year.

ChatGPT Work, which OpenAI introduced in July, is an enterprise-oriented AI agent that companies can put in front of teams for a range of workflows rather than a single chat box. Uber's first exposure to it was a pilot involving 1,000 employees drawn from departments including finance, operations and product. According to Sumanth Sukumar, Uber's vice president of engineering, about 80% of those early adopters were non-engineers — a detail that matters for how the tool is likely to be used company-wide, since it says something about who finds these agents useful beyond the software org.

Sukumar described the kinds of work those pilot users did with it: analyzing complex operational data to spot trends and build recommendations, localizing training materials for different markets and languages, and constructing personalized interactive dashboards for managing day-to-day work. Those are tasks that sit awkwardly between spreadsheets, internal wikis and BI tools today, which helps explain why a general-purpose agent drew interest from finance and ops staff rather than just engineers.

What Uber has not settled is exactly how the rest of its employees should use the product. Sukumar framed the rollout as deliberately open-ended, saying the company wants to make capable AI tools available broadly while leaving teams space to experiment. As an example of that philosophy, he pointed to an earlier internal program in which 30 AI engineers were embedded with teams across Uber for two weeks with a mandate to automate meaningful parts of their jobs. Sukumar said those "AI pods" demonstrated the value of pairing technical skill with deep familiarity with a particular function — an argument, in effect, that the highest-leverage automation comes from embedding specialists rather than issuing a tool and hoping adoption follows.

Uber declined to say what the expanded deployment will cost. Instead, Sukumar said the company intends to monitor usage: how employees apply the technology, what kinds of workflows they build, and whether it makes them more effective. Those findings will inform how Uber deploys AI across the business going forward. Uber also said OpenAI worked closely with it on integrating the chatbot, which suggests a level of customization beyond simply provisioning seats — though neither company has described the technical shape of that integration.

For context, enterprise agents are a relatively new category. Where a standard chat assistant answers questions, an agent-style product is typically designed to carry out multi-step tasks, call into tools and data sources, and produce artifacts such as reports or dashboards. That ambition is also where the risk sits: agents that touch operational data need permissioning, audit trails and some way to check outputs, and localization and dashboard-building both produce work that people will act on. Uber has not described its governance setup, so how much of that burden falls on IT versus individual teams is unknown.

The announcement lands as OpenAI pushes hard into enterprise sales against Anthropic, its main rival. OpenAI built its reputation largely on consumer ChatGPT but has moved quickly to win business customers, and large reference deployments like this one function as marketing as much as revenue — a 30,000-seat rollout is a signal to other CIOs that the product scales. Whether it is genuinely stickier than competing offerings is a question Uber's own usage tracking may eventually answer, at least internally.

Why it matters: A marquee company moving from a 1,000-person pilot to roughly 30,000 seats is a meaningful endorsement for enterprise agent products, and it gives OpenAI a reference customer it can cite in competitive deals. For product and engineering leaders, the notable part is the adoption pattern — most pilot users were non-engineers doing analysis, localization and dashboard work — which suggests the near-term value of these tools may sit in operations rather than software development, and that measuring outcomes will matter more than seat counts. Uber's decision to track workflows rather than publish a cost figure is, by inference, a sign that the return on these deployments is still being figured out across the industry.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

The ride-share giant previously deployed a much smaller pilot for ChatGPT Work but is now taking the enterprise agent company-wide.