Uber Expands ChatGPT Work to 30,000 Employees: What Businesses Can Learn

Uber is moving workplace AI from a limited experiment to a company-wide system. On 22 October 2026, OpenAI announced that ChatGPT Work would become available to more than 30,000 Uber employees worldwide, up from an initial pilot involving 1,000 people.

The headline number is large, but the more useful story is how Uber is approaching adoption. The rollout combines the AI agent with approved integrations, an internal skills library and dedicated training. That makes this less about giving every employee a chatbot and more about building a supported operating layer for AI-assisted work.

What Uber actually announced

According to OpenAI’s announcement dated 22 October 2026, ChatGPT Work is being deployed inside ChatGPT Enterprise as an agent for complex, multi-step knowledge work. It can connect with tools used across the company, including Google Drive, Gmail, Google Calendar and Slack.

The stated use cases include researching information, preparing meeting briefs, creating presentations, analysing data and completing workflows across applications. Uber is also giving teams access to OpenAI’s frontier models, with GPT-6.1 Sol named as the default capability for this rollout. Operations, finance, marketing and product teams are included, while developers will continue using Codex for software work.

This distinction matters. The rollout is not limited to coding, and it is not presented as a customer-facing feature. It is an internal work platform meant to help different functions handle tasks that usually require moving between multiple apps and information sources.

From a 1,000-person pilot to the whole company

TechCrunch reported on 22 October 2026 that around 80% of the pilot users were non-engineers. Employees used the system to analyse complex operational data, localise training material for different markets and languages, and build personalised dashboards for daily work.

Uber also paired 30 AI engineers with teams across the business for two weeks to automate parts of their jobs. That is a useful detail for founders and agency leaders. People who understand a department’s work often know the problems but not how to turn them into reliable AI workflows. Technical specialists can bridge that gap, then document the useful patterns for wider adoption.

Uber’s internal AI skills library appears to serve the same purpose at scale: successful methods do not remain isolated inside one person’s prompts. They can be reused, governed and improved across the company.

Why this matters beyond Uber

Many businesses are still treating generative AI as an individual productivity tool. One employee uses it for summaries, another for copy, and somebody else experiments with data analysis. That can save time, but it rarely changes how the organisation works.

Uber’s approach points towards a more structured phase of enterprise AI adoption. The tool is connected to company systems, employees receive training, integrations are approved, and repeatable skills are collected in one place. For a smaller business, the same idea can be applied without a 30,000-person rollout.

Start with one measurable workflow in a team that feels the problem every week. Give the tool access only to the information it genuinely needs. Define who reviews the output, record the working process and track whether it improves speed, quality or cost. This is the practical core of AI implementation, regardless of company size.

If you are still deciding where AI fits in your organisation, this guide to AI implementation for business explains how to begin with real use cases. A separate AI roadmap guide can help connect those early experiments to longer-term priorities.

What is still missing

The announcement is important, but it is not proof of a return on investment. Uber and OpenAI did not disclose the price of the rollout, quantified productivity gains, error rates or savings from the pilot. Uber said it plans to study how employees use the technology, which workflows they develop and how the system affects their work.

There is also limited public detail about permissions, audit controls, data boundaries or how high-stakes outputs will be checked. Approved integrations are a positive signal, but an agent that can work across email, calendars, files and chat needs careful access control. Businesses should not interpret company-wide availability as permission for an agent to act everywhere.

The 1,000-person pilot may also have attracted unusually enthusiastic users. Scaling from early adopters to the average employee is where many workplace technology projects slow down. Continued usage, reliable outputs and measurable business impact will matter more than the initial number of seats.

A practical rollout lesson for smaller teams

The clearest lesson is that access alone is not an AI strategy. A useful rollout needs three supporting layers: approved tools and data connections, reusable skills or workflows, and training that shows employees when to use the system and when human review is required.

For an agency, that could mean piloting an AI agent on campaign research or reporting rather than every client process at once. For a founder-led business, it might be weekly operations analysis or meeting preparation. The workflow should be frequent enough to measure and low-risk enough to improve safely.

Set the measurement before expanding. Compare turnaround time, correction rate and team effort against the old process. Keep human approval for financial, legal, client-facing or brand-sensitive work. Document the workflow so the value does not depend on one enthusiastic employee or one vendor.

The takeaway

Uber’s ChatGPT Work rollout is a strong signal that AI agents are moving beyond small technical teams and into everyday knowledge work. The significant part is not only the 30,000-employee scale. It is the combination of a pilot, cross-functional support, approved integrations, shared skills and training.

The results are not yet public, so it would be premature to call the rollout a proven success. Still, the operating model is worth studying. Businesses that treat AI as a managed capability, with clear access, reusable workflows and measurable outcomes, are more likely to create lasting value than those that simply buy licences and hope employees figure it out.