Overview
- Uber disclosed in coverage published Sunday that it handpicked about 30 AI-proficient engineers and ran 16 two-week 'Agentic Pods' across functions, and the company reported cuts such as financial planning from 15 hours to 30 minutes.
- Each pod spends the first days shadowing staff to map steps, then rapidly builds and tests AI software in-context with domain experts rather than delivering a standalone product.
- CTO Praveen N. Naga says the biggest gains come from redesigning entire processes—removing redundant approvals, replacing legacy tools, and enabling faster decisions—rather than only speeding single tasks.
- Uber frames the effort as a disciplined response to earlier high AI token spending this year, shifting focus from costly external model use to internal, outcome-driven deployments.
- If scaled, the approach could cut routine work across finance, marketing and HR, but it will require clear metrics for return on investment, changes to staff roles, and continued oversight of model costs and safety.