Uber Goes All-In on OpenAI: How ChatGPT Work Could Transform 30,000 Jobs

9 minutes ago

8 min read

Uber’s decision to deploy ChatGPT Work across its global workforce marks a significant development in the transition from experimental generative AI to enterprise-wide agentic automation. The ride-hailing and delivery company is expanding its collaboration with OpenAI to make the AI platform available to more than 30,000 employees, following an initial pilot involving 1,000 workers.

The announcement highlights a broader shift in corporate artificial intelligence. Businesses are moving beyond using AI primarily to generate text, summarize documents, or answer questions. They are increasingly exploring systems that can coordinate complex tasks, interact with multiple applications, analyze operational data, and produce completed work.

For Uber, the rollout is an opportunity to discover how these capabilities can improve productivity across engineering, finance, operations, marketing, and product development. For the wider technology industry, it provides a prominent example of how AI agents could become part of everyday enterprise infrastructure.

What Is ChatGPT Work, and Why Is Uber Deploying It Company-Wide?

ChatGPT Work is an enterprise-focused agentic platform designed to help employees complete multistep professional tasks within ChatGPT Enterprise. Rather than requiring users to manage every stage of a workflow themselves, the system can use connected tools, process information, and help deliver finished outputs.

Uber’s expansion builds on a pilot launched earlier in 2026. That initial deployment included employees from different business functions, giving the company an opportunity to assess how AI could support work beyond software development.

One of the pilot’s most important findings was the participation of non-engineers. According to Uber Vice President of Engineering Sumanth Sukumar, approximately 80% of early adopters were employees outside engineering.

This matters because enterprise AI adoption is often associated with coding assistants and technical teams. Uber’s experience suggests that many valuable applications may emerge in departments where employees spend significant time interpreting information, coordinating activities, preparing reports, and making operational decisions.

The company is not prescribing one universal use case. Instead, it is making the technology broadly available and allowing teams to experiment with workflows suited to their responsibilities.

From 1,000 Pilot Users to More Than 30,000 Employees

Scaling an AI experiment across an organization is fundamentally different from testing a tool with a small group of enthusiastic employees.

A limited pilot can identify promising applications, but a company-wide deployment must accommodate different skill levels, business requirements, data-access permissions, and working practices. It also needs a framework for evaluating whether the technology creates measurable value.

Uber’s rollout represents an expansion from an initial group of 1,000 employees to a workforce exceeding 30,000. This gives the company a much larger environment in which to identify productive use cases and understand how agentic AI performs across different departments.

During the pilot, employees used AI to analyze operational datasets, identify patterns, develop recommendations, localize training materials for different markets and languages, and create interactive dashboards tailored to their work.

These examples demonstrate three important categories of enterprise AI value:

Information analysis: Turning complex datasets into patterns, explanations, and actionable recommendations.
Workflow acceleration: Reducing the manual effort required to prepare documents, reports, presentations, and training materials.
Personalized productivity: Building tools and dashboards that help employees monitor responsibilities and make decisions more efficiently.

The next challenge is determining which of these applications produce meaningful, repeatable improvements at organizational scale.

How ChatGPT Work Can Transform Enterprise Workflows

The significance of an AI agent depends on more than its ability to generate a high-quality answer. Its value increases when it can connect information, perform intermediate steps, use appropriate tools, and help deliver a complete result.

ChatGPT Work connects with workplace applications including Google Drive, Gmail, Google Calendar, and Slack. These integrations can help employees bring information from different systems into a unified workflow.

Consider a manager preparing for an important business review. Traditionally, the process might involve searching email threads, locating reports in shared drives, reviewing calendar details, collecting updates from colleagues, and assembling a presentation.

An agentic system can help coordinate these steps by retrieving relevant information, synthesizing the findings, preparing a meeting brief, and organizing the material for presentation.

Similar workflows can support finance teams reviewing performance, operations teams investigating service disruptions, marketing teams preparing campaigns, and product teams evaluating customer feedback.

The critical distinction is that an agent can help manage the process, rather than simply provide information about how to complete it.

However, these workflows still require appropriate permissions, accurate source data, and human review where decisions carry significant consequences. Automation is most effective when tasks have clear objectives and outputs that employees can verify.

Why Non-Engineering Teams Could Benefit Most

Uber’s early adoption figures challenge the assumption that enterprise AI primarily benefits developers.

Software engineers can use AI to accelerate coding, testing, and debugging. Yet many non-technical employees perform work that also involves repeatable information-processing tasks.

Finance professionals analyze spreadsheets, reconcile figures, investigate discrepancies, and prepare forecasts. Operations teams evaluate performance across markets and identify emerging problems. Human resources teams coordinate hiring, onboarding, and training. Marketing departments research audiences, develop campaign materials, and adapt content for different regions.

These activities often involve moving between multiple applications and converting unstructured information into decisions or deliverables.

AI agents could reduce this administrative burden, allowing employees to spend more time on judgment, collaboration, and specialized expertise.

The benefit will not be uniform. Tasks requiring nuanced negotiation, sensitive interpersonal decisions, or highly accountable professional judgment may remain difficult to automate reliably. The strongest early applications are likely to be workflows where the desired outcome is clear, the information is accessible, and errors can be detected before they cause significant harm.

AI Pods: Combining Technical Expertise With Business Knowledge

Uber has also experimented with a collaborative model in which AI specialists work directly with employees who understand particular jobs.

Earlier in 2026, the company paired 30 AI engineers with teams across Uber for two-week periods, challenging them to automate important parts of their work. These groups, described as AI pods, were intended to combine technical capability with practical knowledge of specific business functions.

This approach addresses a recurring problem in enterprise AI adoption: the gap between what a technology can theoretically accomplish and what an organization actually needs.

An engineer may understand how to build an automated workflow, but an operations specialist knows which exceptions matter, where data quality problems arise, and which decisions require escalation. Bringing these perspectives together can produce more useful systems than deploying a generic assistant without adapting it to real working conditions.

The model also provides a way to discover opportunities that employees might not identify through conventional training alone.

As ChatGPT Work becomes available across Uber, this kind of collaboration could help transform individual experiments into repeatable internal practices.

Frontier AI Models, Multimodal Capabilities, and Codex

The rollout gives Uber access to OpenAI’s portfolio of frontier models, with GPT-6.1 Sol identified in the announcement as the default capability within ChatGPT Work. The deployment is designed to balance performance with cost efficiency.

Model selection is increasingly important for enterprise applications because different tasks impose different requirements. A straightforward summary does not necessarily require the same computational resources as a complex financial analysis or a workflow involving several applications.

OpenAI’s multimodal capabilities, including voice and image generation, provide additional ways for employees to interact with AI and produce work. Multimodal systems can be particularly useful when information is distributed across written documents, visual material, and spoken communication.

Uber will also continue using Codex for software engineering and development workflows. This distinction reflects an emerging enterprise strategy: using different AI products or capabilities for different categories of work rather than expecting one interface to serve every purpose equally well.

The rollout will be supported by an internal AI skills library, approved integrations, and employee training. These components are important because access to a powerful model does not automatically give every employee the knowledge to use it effectively.

How Uber Will Measure the Business Value of AI

Uber has not disclosed the cost of the company-wide deployment. Instead, it has indicated that it will examine how employees use the platform, which workflows they develop, and whether those workflows help them work more effectively.

This approach places measurement at the center of the rollout.

Counting prompts, active users, or generated documents can show whether a tool is being used, but these metrics do not establish whether it is improving business performance. A more meaningful assessment must connect AI activity to the outcomes that matter for individual roles and the wider organization.

Potential evaluation criteria include:

Measurement area	What Uber could evaluate
Productivity	Time required to complete comparable tasks
Work quality	Accuracy, completeness, and usefulness of deliverables
Workflow efficiency	Number of manual steps eliminated or simplified
Adoption	Sustained use across departments and job functions
Reliability	Frequency of errors, rework, and human intervention
Business impact	Improvements in operational decisions and measurable outcomes

These are useful evaluation categories, not confirmed metrics Uber has committed to publishing.

Establishing credible results will require baselines, consistent measurement methods, and attention to unintended consequences. For example, a workflow that produces a report faster may still create additional work if employees must spend substantial time checking errors.

The objective should be improved outcomes, not merely increased AI activity.

The Risks of Enterprise-Wide AI Agent Adoption

Deploying AI throughout a large organization introduces challenges that smaller experiments may not reveal.

Data security and permissions: Connected applications can contain confidential business information, employee records, financial data, and customer communications. Access controls must ensure that employees and AI systems retrieve only information they are authorized to use.

Accuracy and verification: AI-generated analysis can contain mistakes or overlook important context. Financial calculations, operational recommendations, and management reports need appropriate validation.

Workflow reliability: Multistep tasks create more opportunities for errors than isolated text generation. A mistake early in a workflow can affect every subsequent stage.

Employee training: Workers need to understand how to delegate tasks, assess outputs, recognize uncertainty, and handle sensitive information responsibly.

Cost and return on investment: Enterprise deployments introduce licensing and implementation expenses, as well as the cost of training, integration, oversight, and workflow redesign.

Governance and accountability: Organizations need clear rules about which actions AI may perform independently, which require approval, and who is responsible for the final outcome.

These challenges do not negate the benefits of agentic AI. They explain why a successful rollout requires organizational design and governance alongside technical capability.

OpenAI's Enterprise Strategy and the AI Platform Competition

Uber’s decision also highlights the intensifying competition among AI companies to become essential providers of workplace technology.

OpenAI has been expanding its enterprise business, while Anthropic and other competitors are pursuing their own corporate AI offerings. Competition increasingly extends beyond model quality to include integrations, security, administrative controls, specialized agents, and the ability to complete useful work across applications.

For enterprise customers, the long-term decision may depend on how well an AI platform fits existing infrastructure, protects data, supports employees, and delivers measurable value.

Large deployments such as Uber’s can also provide valuable feedback about real-world usage. Diverse employees encounter a wider range of tasks and edge cases than a small group of specialists, helping organizations identify where AI works well and where additional safeguards or customization are necessary.

The broader industry question is whether agentic platforms can become reliable enough to serve as a standard layer between employees and the software they use every day.

What Uber's ChatGPT Work Rollout Means for the Future of Work

Uber’s deployment of ChatGPT Work across more than 30,000 employees represents a major test of enterprise AI at organizational scale. It builds on a 1,000-person pilot that demonstrated potential applications beyond engineering, including operational analysis, localized training, and personalized dashboards.

The rollout's ultimate significance will depend on what Uber learns about productivity, workflow design, adoption, reliability, and cost. Broad access creates opportunities, but sustained business value requires employees to use AI in ways that improve actual outcomes.

For other organizations, the key lesson is that successful AI adoption is not simply a software procurement decision. It involves identifying suitable tasks, combining technical expertise with domain knowledge, training employees, measuring results, and establishing clear controls over sensitive work.

As Dr. Shahid Masood and the expert team at 1950.ai examine the evolution of artificial intelligence and enterprise technology, Uber’s decision offers an important example of the shift toward agentic systems. The next stage of corporate AI will be shaped not only by more capable models, but by how effectively organizations integrate those models into everyday operations.

The central question is no longer just whether employees can use AI. It is whether AI can help entire organizations complete meaningful work more efficiently, consistently, and responsibly.