Uber says its AI engineers are learning how employees work to build AI tools for HR and other departments

Uber is embedding AI engineers inside teams like finance, HR and marketing to understand how employees work before building AI tools. The company says redesigning entire workflows with AI has reduced processes that once took days to just minutes.

In Short

• AI engineers now shadow employees before building AI solutions
• Uber says some workflows now take minutes instead of days
• CTO says AI delivers bigger gains by redesigning work, not just speeding up tasks

Uber is taking a different approach to using AI inside the company. Instead of simply asking employees to use AI tools, the ride-hailing giant is first trying to understand how people work and then redesigning those workflows around AI. The company's Chief Technology Officer, Praveen Neppalli Naga, has revealed that Uber is sending its top AI engineers into teams such as finance, legal, marketing, customer support, human resources, and procurement to observe how employees perform their day-to-day work closely. The goal is to identify bottlenecks and build AI-powered systems that improve entire workflows rather than just speeding up individual tasks.

Uber has created dedicated AI teams that work directly with different business units for short, focused projects. Known internally as "Agentic Pods," these groups spend about two weeks inside departments, first learning how employees complete their day-to-day work before developing AI-powered solutions tailored to those processes, Business Insider reported.

The company says this hands-on approach has led to dramatic improvements. A financial planning task that once took nearly 15 hours now takes around 30 minutes, while preparing financial reports has been reduced from two days to roughly 10 minutes. In the marketing division, quality assurance checks that previously stretched over two weeks can now be completed in under an hour.

Naga believes the real value of AI is not in making individual tasks slightly faster. Instead, he says companies can unlock much bigger productivity gains by redesigning entire business processes. That includes removing unnecessary approval chains, replacing legacy software with AI-driven systems, and enabling employees to make decisions more quickly.

A new way of deploying AI inside companies

Industry executives have likened Uber's initiative to the "forward-deployed engineer" model that has become increasingly popular in Silicon Valley, where engineers work closely with customers to understand their needs before building AI-powered products. Uber, however, is applying the same idea internally. Rather than working with external clients, its AI engineers are partnering with teams across the company to identify inefficient processes and rebuild them using AI.

Peter Wilczynski, chief product officer at Vantortech, humorously described this role as a "Rearward Deployed Engineer," saying these engineers are embedded within corporate functions to rethink entire workflows instead of merely accelerating individual tasks.

Naga's remarks also come a few months after Uber attracted attention during the industry's "tokenmaxxing" debate, which questioned whether rising AI usage was translating into meaningful productivity gains.

Earlier this year, Uber executives drew attention after discussing the company's growing AI investments. In May, Uber COO Andrew Macdonald said the company had not yet found a clear link between higher AI token usage and measurable productivity gains.

"That link is not there yet, right?" Macdonald said. "I think maybe implicitly there is more that is getting shipped, but it's very hard to draw a line between one of those stats and, 'OK, now we're actually producing 25% more useful consumer features.'"

The discussion is not surprising because companies across the tech industry significantly increased spending on AI tools. Reports at the time suggested Uber had already exhausted its planned AI budget for 2026 within the first four months of the year, suggesting how rapidly enterprise AI costs are rising.

- Ends