Most AI-in-the-workplace stories end with a headcount. Uber’s newest one, announced on 22 October, is interesting because it ends with a head start instead: the company is putting ChatGPT Work in front of every one of its roughly 30,000 employees, scaling up from a 1,000-person pilot run earlier this year, as reported by TechCrunch.
🔍 THE BOTTOM LINE: This is one of the largest corporate agent deployments to date, but the details that matter for careers are the quieter ones. Eighty percent of the pilot’s early users were non-engineers. Uber paired 30 AI engineers with business teams for two-week “AI pods” — a formal programme of staff coaching staff. And when asked what success looks like, the company’s answer was adoption and workflow quality, not cost savings. That’s a company betting that the hard part of AI at work isn’t the software; it’s the 30,000 people learning to use it.
What Uber actually announced
ChatGPT Work — OpenAI’s enterprise agent product, which landed in July — will be available company-wide, integrated with the tools employees already use, as The Next Web’s coverage confirms. Uber’s VP of engineering Sumanth Sukumar told TechCrunch the earlier pilot reached workers across finance, operations and product, and that a majority of early adopters — about 80 percent — were non-engineers. They used it for analysing operational data, localising training materials across markets and languages, and building personal dashboards for their daily work.
Uber’s software engineers stay on a separate development platform, so this rollout is pointed squarely at the non-coding majority of the workforce. That detail matters for any worker trying to guess where AI lands first: the first wave here isn’t engineering, it’s everything around engineering.
The training programme hiding in the announcement
The most career-relevant line in the coverage is easy to skim past. Sukumar described “AI pods” from earlier this year: 30 AI engineers paired with teams across Uber, given two weeks to automate key parts of the team’s job. His framing of the result — that the opportunity shows up “when you combine technical expertise with a deep understanding of a particular job” — is a description of exactly the skill blend the survey data says employers are struggling to find, as Harvard Business School’s research coverage put it: firms aren’t training people formally, so employees use AI “like glorified Google” and get little out of it.
Uber is also declining to promise a productivity number. Sukumar told TechCrunch the company will track how employees use the tool and steer later deployment from what it learns. CTO Praveen Neppalli Naga framed the goal as letting agents handle “operational friction” so teams can focus on “high-value problems that only humans can solve,” as Forkast’s write-up of the OpenAI announcement reports. Uber’s own NZ-market blog has documented the company’s workplace AI experiments since the pilot days.
Worth holding onto: Uber’s earlier reporting on its own AI budget was less rosy — this site covered in September how Uber burned through its AI budget in four months and struggled to justify the spend. A company-wide rollout with no productivity target attached is either confidence or a measurement problem — possibly both.
What it means for workers
Two signals sit inside this rollout for anyone watching their own job. First, the deployment is enterprise-wide by design — 30,000 seats, not a 10 percent engineering vanguard — which treats AI fluency as a baseline job skill rather than a specialism. Second, the mechanism Uber chose for spreading it was internal coaching: engineers embedded with operations teams, structured two-week sprints, adoption measured before savings. That’s close to a case study of what “reskilling” actually looks like when a company funds it properly.
For NZ readers, the transferable part isn’t Uber. It’s the skill pattern Sukumar described: domain depth plus the ability to wire AI into that domain. The workers most exposed in the displacement data have been juniors in codified-knowledge jobs — a pattern this site tracked through Stanford’s entry-level research — while people who sit between the technology and a specific business problem keep gaining leverage. A two-week pod is one company’s answer to building exactly that.
FAQ
What is ChatGPT Work? OpenAI’s enterprise agent product launched in July 2026, aimed at workplace workflows rather than consumer chat. It can run multi-step tasks across the tools a company already uses.
Are companies replacing workers with ChatGPT Work? Uber’s announcement frames the rollout around adoption and workflow building, with no headcount target attached, and 80 percent of its pilot users were non-engineers. The company says engineers stay on separate tooling.
What are “AI pods”? An Uber programme pairing 30 AI engineers with business teams for two-week sprints to automate parts of the team’s work — an internal coaching model Uber credits for showing what AI can do in a specific job.
What should workers learn from Uber’s rollout? The skills Uber concentrated on are domain knowledge plus workflow building — analysing operational data, localising materials, constructing dashboards. The pattern matches broader employer surveys finding formal AI training is rare and valuable.
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
