Uber's AI chief on why the 'debate is over' for adoption

Uber’s Jai Malkani says the time for debating AI adoption has passed. As the company’s Global Head of Product for Customer Obsession, he’s deploying AI across a logistical beast that coordinates riders, drivers, eaters, couriers, restaurants, and grocery shoppers—often in the same app. The complexity is staggering, and the AI strategy is ruthlessly segmented by product. For ride fare disputes, models interpret route changes from traffic or tolls to offer equitable resolutions. For a wrong Eats order, AI analyzes photos and tracks an item’s lifecycle to figure out exactly where the kitchen screwed up. For grocery, it mediates real-time inventory nightmares between shoppers and consumers.

But Malkani’s real focus is on the workforce, where he frames AI as an ‘intelligent co-pilot’ rather than a replacement. Customer service agents get conversational summaries, automated investigations, and what he calls ‘empathetic next-best responses.’ The idea is to turn a complex policy document into an actionable routine for a human agent in seconds—saving them from the drudgery while supposedly improving the quality of care. It’s a productivity play for everyone from developers to operations.

Hyper-personalization is the other pillar. Malkani points to a specific example that’s less about flashy consumer features and more about preventing operational meltdowns: the system pings drivers before documents expire, tailoring requirements to their specific region, account type, and trip history. It’s the unsexy plumbing of gig work, but getting it wrong means drivers get kicked offline. Getting it right means the platform hums.

They’re measuring success through a mix of traditional customer experience scores, resolution speed, and a metric that tells the real story—incremental gross bookings across geographic segments. Malkani’s advice to other leaders boils down to a call to action: get intimate with the tech, hunt down high-value use cases, lock in executive sponsorship, and show the money. For a company that still reports billion-dollar losses some quarters, that last bit of advice probably isn’t optional.