Uber Puts All 30,000 Employees on ChatGPT Work: What the Rollout Really Means

Key Takeaways

• Uber and OpenAI announced on Thursday 22 October 2026 that Uber’s entire global workforce of about 30,000 people will now use ChatGPT Work, expanding an earlier pilot of 1,000 employees.
• ChatGPT Work is an enterprise-focused AI agent that corporate teams can deploy for multi-step workflows, introduced in July according to TechCrunch.
• About 80 percent of the pilot participants were non-engineers, Uber vice president of engineering Sumanth Sukumar told TechCrunch, with finance, operations, and product teams among those included.
• Pilot uses included analysing complex operational data, localising training materials for different markets and languages, and building personalised interactive dashboards.
• Uber has not settled exactly how employees will use the tool, and it did not disclose the cost of the rollout, so pricing and company-wide productivity effects remain unconfirmed.
• Company-wide availability is not the same as company-wide adoption: the announcement describes who can use the tool, not proof that every employee already uses it or works measurably faster.

Uber has gone all in on workplace AI. On Thursday 22 October 2026, Uber and OpenAI announced an expansion of their partnership, with Uber telling TechCrunch that its entire global workforce, some 30,000 people, will now be using ChatGPT Work. The move scales up an earlier pilot of 1,000 employees to the whole company, and it is one of the clearest signals yet that enterprise AI has moved from experiment to standard equipment.

The announcement deserves careful reading, because its most interesting detail is also its most honest one. Uber has not settled exactly how employees will use the tool. Rather than prescribing a fixed set of use cases from the top down, the company says it is giving teams room to experiment and will watch what workflows emerge. That approach, part rollout and part research project, tells us as much about the current state of workplace AI as the headline number does.

What ChatGPT Work Is

ChatGPT Work is an enterprise-focused AI agent that corporate teams can deploy for various workflows, introduced in July according to TechCrunch. The distinction between a chatbot and an agent matters here. A chatbot answers a question and stops. An agent carries out multi-step work: gathering information, using approved tools, producing reviewable files, and completing a task that would otherwise involve several separate steps by a human.

That distinction explains why Uber’s pilot results look the way they do. The tasks Sukumar described, analysing operational data to identify trends and develop recommendations, localising training materials, and building interactive dashboards, are not single questions. Each one involves collecting material, transforming it, and producing something another person can use. An agent that can work through those steps, rather than merely drafting text about them, is a different kind of workplace tool from a chat window.

It is worth noting what the coverage does not specify. The reporting reviewed for this article does not publish a feature list for ChatGPT Work, does not describe its data controls in technical detail, and does not state which internal Uber systems it can access. What is confirmed is the shape of the product, an enterprise agent for team workflows, and the shape of the rollout, from a 1,000-person pilot to a 30,000-person deployment.

For readers new to this generation of tools, our guide to GPT-6 on every ChatGPT plan explains how ChatGPT’s interface and capabilities are changing, and why agent-style features are becoming central to how people work with AI rather than a side feature.

From a 1,000-Person Pilot to the Whole Company

Uber originally piloted ChatGPT Work with 1,000 employees earlier in 2026, and the composition of that pilot is the most revealing number in the story after the headline itself. About 80 percent of the early adopters were non-engineers, according to Sukumar. The pilot drew workers from across Uber’s departments, including finance, operations, and product.

That mix challenges a common assumption about workplace AI: that it is mainly a tool for software developers. Developers were a minority of the pilot. The majority were people whose daily work involves data, documents, coordination, and communication, the kind of work that fills most large companies. If an AI agent proves useful mainly to engineers, its company-wide value is limited. If it proves useful to finance analysts, operations managers, and product teams, its value is far broader, and Uber’s pilot was designed to test exactly that question.

The tasks those pilot users performed point in the same direction. Analysing complex data from operations to identify trends and develop recommendations is analytical work, not code. Localising training materials for different markets and languages is communication work across a global company. Building personalised interactive dashboards to manage daily work is a mix of analysis and organisation. None of these is a coding task, and all of them are the kind of work that consumes hours in a large organisation without ever appearing in a product release. Uber also ran a smaller experiment alongside the pilot, pairing 30 AI engineers with teams across Uber for two weeks to automate key parts of their jobs. Sukumar called these AI pods, and said they showed how much opportunity there is when technical expertise is combined with a deep understanding of a particular job.

Why an Enterprise-Wide Rollout Matters

A company-wide rollout matters for reasons that go beyond Uber itself. First, it is a vote of confidence that is hard to fake. Giving 30,000 employees access to a tool is a substantial commitment of money, training effort, and IT support, even though Uber did not disclose the cost. Companies do not make that commitment for a product they expect to abandon in a quarter.

Second, it changes the unit of adoption. Most workplace AI stories so far have been about individual enthusiasts or single teams. A whole-company rollout tests something different: whether an agent can be useful across roles, regions, and skill levels at once. The results, if Uber shares them, will be more informative than pilot anecdotes, because they will reflect ordinary employees rather than early adopters.

Third, the rollout signals a shift in how vendors and buyers relate. OpenAI worked closely with Uber on the integration, according to the coverage, which suggests that enterprise AI deals are becoming partnerships rather than simple software purchases. Smaller companies will not get the same level of vendor attention, but they can learn from the pattern: budget for deployment support and discovery time, not just licences.

What Remains Unconfirmed

Honest coverage of this announcement must mark several open questions. Uber did not disclose the cost of the rollout, so any figure attached to it elsewhere should be treated as speculation unless Uber or OpenAI confirms it. No deployment timeline beyond the announcement itself has been published in the coverage reviewed, so readers should not assume that all 30,000 employees received access on the same day or completed training by any particular date. No productivity measurements have been published either: the announcement describes availability, not proof that employees work measurably faster or better with the tool. Uber says it will monitor how employees use the technology, what workflows they develop, and whether it helps them work more effectively, and will use those findings to guide its broader AI deployment. Until those findings appear, productivity claims are unconfirmed.

Data and access questions are also open. The coverage does not state which Uber systems ChatGPT Work can reach, what data controls govern its use, or how the company handles sensitive operational information inside agent workflows. Those details matter enormously for any company considering a similar rollout, and they cannot be inferred from the announcement alone. They would need to come from Uber or OpenAI directly.

Finally, the 30,000 figure itself deserves precision. TechCrunch describes it as Uber’s entire global workforce, some 30,000 people. That is the population gaining access, not a count of daily active users, and it does not include Uber’s drivers, who are not employees. Conflating the workforce figure with driver numbers, or with active usage, would misrepresent the announcement.

What Other Companies Can Learn

For companies watching Uber’s move, five practical lessons stand out. First, pilot beyond engineering. Uber’s most useful finding may be that 80 percent of its pilot users were not engineers, which suggests that testing an agent only with developers understates its potential audience. Second, pair technical skill with job knowledge, as Uber’s AI pods did, rather than expecting either side to discover valuable workflows alone. Third, allow room for experimentation, but measure it: Uber’s plan to monitor usage, workflows, and effectiveness is the part of the announcement other companies should copy most carefully, because unmeasured experimentation becomes untraceable spending. Fourth, treat rollout as deployment plus discovery, and budget time for both. Fifth, be honest about what is unproven, including cost, timeline, and productivity, when making the internal case for a similar investment.

It is also worth placing this rollout in the wider workflow picture. An agent that analyses data and builds dashboards is most valuable inside a company that already has clear workflows for it to improve, which is why our guide to building a scalable content workflow is relevant even outside media: the discipline of defining a workflow before automating it applies to operations, finance, and product teams just as much as to content teams.

The Bigger Picture for Workplace AI

Uber’s rollout arrives at a moment when workplace AI is crossing from novelty to infrastructure. The question for most companies is no longer whether AI tools exist, but whether they can be deployed safely, usefully, and measurably across an entire workforce. Uber’s answer, for now, is to deploy broadly, let teams experiment, and study the results. That answer is pragmatic rather than triumphal, and it is more credible for it. If the experiment succeeds, expect other large employers to follow the same pattern. If it stalls, the reasons will be instructive too. Either way, the most important numbers in the story have not been published yet: how employees actually use the tool, what they build with it, and whether it genuinely helps them work better.

Sources

• TechCrunch, “Uber puts all 30,000 employees on ChatGPT Work”, 22 October 2026: https://techcrunch.com/2026/10/22/uber-puts-all-30000-employees-on-chatgpt-work/
• Mezha, “Uber expands ChatGPT Work to all 30,000 employees”, 22 October 2026: https://mezha.net/eng/news/mezha-9fdd22fa30fbdf82a9dd441c_uber_expands_chatgpt/