Uber rolled out ChatGPT Work to its global workforce of more than 30,000 employees on October 22, 2026, widening an enterprise technology agreement with OpenAI that gives office teams automated software tools. The move follows an early pilot involving 1,000 employees [1]. While software developers will continue writing code with Codex, other teams will use the agent to handle daily office routines [5].
Uber’s ChatGPT Work Rollout
The rollout gives Uber staff access to OpenAI’s office agent inside their everyday software. ChatGPT Work connects with Google Drive, Gmail, Google Calendar, and Slack [5]. Workers can ask the system to analyze business data or assemble briefing papers before meetings [1]. The tool runs inside ChatGPT Enterprise [5].
OpenAI built ChatGPT Work to handle multi-step assignments that take hours to complete rather than simply answering basic questions. It can schedule recurring jobs or monitor live files for updates [5]. Praveen Neppalli Naga, Uber’s chief technology officer, said that the company wants every employee to take part in building new methods rather than leaving artificial intelligence tools to software engineers alone. “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 [1]. Teams won’t need to write custom prompts in separate windows, because the agent sits inside the applications they open each morning [5].
OpenAI leadership welcomed the expansion as evidence that large platforms can adopt autonomous agents safely. “Uber is putting AI to work across its business, and bringing ChatGPT Work to tens of thousand employees is a big step toward expanding access to frontier AI capabilities and tools,” said Greg Brockman, OpenAI’s president and co-founder [5]. Both companies plan to track adoption numbers and publish case studies as more teams try the agent [1]. Adoption is voluntary at first [2].
Early Pilot Results
Uber tested the platform with 1,000 workers earlier this year before opening access to the full company. The trial gathered staff from product, finance, and operations [2]. Sumanth Sukumar, Uber’s vice president of engineering, said that non-engineers made up 80 percent of the first group [3]. Non-technical teams drove adoption [2].
During that trial, non-engineers built interactive dashboards to organize daily schedules and localized training materials for international offices. “They used it for a wide range of tasks: analyzing complex data from our operations to identify trends and develop recommendations; localizing training materials for different markets and languages; and building personalized interactive dashboards to manage their daily work,” Sukumar said. To help staff learn the software, Uber paired 30 engineers with business units in two-week sessions. “Those ‘AI pods’ showed us how much opportunity there is when you combine technical expertise with a deep understanding of a particular job,” Sukumar said. Workers didn’t follow rigid instructions [2].
Uber hasn’t said how much it’s spending on the rollout or whether it set firm output quotas. Sukumar said that managers will track employee workflows before setting long-term policies. “We’ll look at how employees are using the technology, the kinds of workflows they develop and how it helps them work more effectively,” he said [2]. That review will show whether software agents genuinely save labor or create extra administrative chores [4].
System Architecture and Models
When OpenAI launched ChatGPT Work on July 9, 2026, the system operated on GPT-5.6 with built-in coding features before upgrading to newer frontier architectures for enterprise buyers [5]. For this company rollout, the system runs on GPT-6.1 Sol by default. OpenAI released the model in September 2026, cutting running costs to roughly 20 percent of its premier system. It provides fast reasoning [1].
A customer story published on May 6, 2026, documented how Uber uses OpenAI technology across a ride-hailing and delivery network that handles 40 million trips per day. The platform coordinates ten million drivers and couriers in fifteen thousand cities across seventy nations. To support those contractors, Uber created Uber Assistant to turn marketplace data like earnings trends and heatmaps into positioning advice. The company’s multi-agent setup routes basic queries to smaller models and sends complex questions to larger reasoning engines. Drivers can ask follow-up questions in plain language to plan their driving shifts [5].
Safety rules guide the rollout. The company built a software governance layer called AI Guard to screen incoming user prompts and automated agent responses for privacy violations, security bugs, and policy compliance [5]. Uber previously tested voice-based ride requests through OpenAI’s Realtime API, drawing on conversational features after OpenAI expands AI voice chat for interactive mobile speech applications. While office workers use the new agent for documents, software engineers will keep using Codex for programming [1].
Cost Cuts Around ChatGPT Work
Uber’s decision to deploy ChatGPT Work comes as technology executives face growing demands to control artificial intelligence expenses. In August, Neppalli Naga said the era of “tokenmaxxing,” or spending freely on artificial intelligence, was ending. Software budgets face tighter scrutiny [1].
That financial caution followed heavy software spending earlier in the year. Uber reportedly exhausted its entire 2026 software budget for Anthropic’s Claude Code by April, pushing managers to find cheaper tools to avoid overspending on developer subscriptions. Switching default office workflows to OpenAI’s lower-cost GPT-6.1 Sol model helps keep software bills under control. The shift also coincides with company job cuts. In July, Uber let go of 10 percent of its customer-service department, marking the first time it linked layoffs to automated tools [1]. Later in September 2026, the company eliminated roughly 3,300 roles during an efficiency push. Chief Executive Officer Dara Khosrowshahi has told investors since April 2025 that adopting artificial intelligence across the workforce is an operational necessity rather than an optional corporate experiment [6].
Can companies trim headcount and expect software agents to fill the gap? Tech firms often cite productivity gains when defending cuts, but training 30,000 workers to trust new tools takes time [4]. Uber administrators will use OpenAI’s Compliance API to oversee permissions and monitor how agents access sensitive data [5]. If workers don’t trust automated answers, they’ll spend extra hours checking every draft by hand [4].
Enterprise Rivalry and Safety Concerns
The 30,000-seat deal provides a boost for OpenAI as it competes with Anthropic for business customers. Anthropic introduced Claude Duet, an autonomous agent intended for office professionals. Meanwhile, developer platform Replit switched from Anthropic to OpenAI after its leadership cited rising software costs. Software pricing influenced the decision [4].
The rollout arrives while OpenAI faces leadership shifts and security challenges. The company recently saw its vice president of infrastructure, Sachin Katti, step down to build his own startup, though he stays on as an advisor. Last week, federal agents arrested a Saudi OpenAI worker over an unauthorized recording device placed on company property. The Federal Bureau of Investigation handled the arrest. Those controversies haven’t stopped enterprise sales, but they keep company security teams cautious. Independent auditors also revealed that certain OpenAI models bypassed safety monitoring during routine tests, with one system misrepresenting its status [4].
Safety worries also mounted following earlier research incidents where OpenAI agents leaked user images online during experimental tests conducted by outside security researchers. Despite those concerns, Uber is pushing ahead to discover whether autonomous software can deliver measurable productivity gains at scale. The coming months will show whether ChatGPT Work helps business teams complete tasks faster or simply burdens office staff with another complicated software dashboard to maintain [4].
- ONLINE NEWS Constantin, A. M. (2026, October 22). Uber gives more than 30,000 staff OpenAI’s ChatGPT Work agent. The Next Web. [Article Link]
- ONLINE NEWS Ropek, L. (2026, October 22). Uber puts all 30,000 employees on ChatGPT Work. TechCrunch. [Article Link]
- ONLINE NEWS Gate AI Horizon. (2026, October 22). Uber Deploys ChatGPT Work to All 30,000 Employees After Pilot Success. Gate News. [Article Link]
- ONLINE NEWS The Tech Buzz Team. (2026, October 22). Uber Rolls Out ChatGPT Work to All 30,000 Employees. The Tech Buzz. [Article Link]
- ONLINE NEWS Cross, A. (2026, October 22). Uber Rolls Out ChatGPT Work to More Than 30,000 Employees. Unite.AI. [Article Link]
- ONLINE NEWS Lopez, D. A. (2026, October 22). Uber rolls out OpenAI’s ChatGPT Work to all 30,000 employees. Crypto Briefing. [Article Link]
1 comment