Uber tests embedded AI ‘Agentic Pods’ model

Uber (UBER) is testing a new way to deploy artificial intelligence by embedding its top AI engineers directly into internal business units. Under chief technology officer Praveen Neppalli Naga, the company has formed short, intensive project teams that work alongside staff in functions including finance, legal, marketing, customer support, human resources and procurement.

These teams, branded as “Agentic Pods,” are designed as two-week engagements. Rather than building generic tools in isolation, the pods operate inside the departments they serve, with engineers working closely with the employees whose workflows they aim to improve.

At the start of each engagement, engineers in the Agentic Pods spend several days shadowing employees to understand existing processes step by step. The goal is to map entire workflows, including handoffs, approvals, and the legacy software in use, before writing any code.

Once they have captured the full workflow, the pods shift into rapid build-and-test cycles. The teams create software that can automate or streamline large portions of the process, then iterate quickly based on feedback from the people performing the work, aiming for tangible changes within the two-week window.

Initial pilots of the Agentic Pod model have produced significant time reductions in multiple areas of Uber’s operations. One reported example is a financial planning workflow that was cut from 15 hours to 30 minutes after the pod redesigned the process and deployed new tools.

In financial reporting, the time required to generate certain reports reportedly fell from two days to about 10 minutes. Marketing teams saw quality checks shrink from roughly two weeks to under an hour once workflows were restructured and supported by new software.

The initiative emphasizes that the biggest efficiency gains come from redesigning end-to-end workflows, not just speeding up isolated tasks. By examining every step, the pods target unnecessary approvals, redundant work and dependence on legacy systems that slow decision making.

This approach positions AI as a lever for organizational change, with engineers empowered to question how work is structured across departments. The focus is on enabling faster, more data-driven decisions while reducing manual effort in routine but complex processes.

Industry observers have compared the Agentic Pod concept to the forward-deployed engineer roles seen at some technology companies, where specialists embed with customers to build tailored solutions. In Uber’s case, the embedding happens within its own operations, focusing on internal efficiency rather than external clients.

One executive has described this internally focused role as a “Rearward Deployed Engineer,” highlighting the shift in where advanced AI talent is placed. Instead of concentrating on outward-facing products alone, Uber is directing these skills toward reengineering its internal workflows as part of its broader AI strategy.