DoorDash acquires YC-backed applied AI lab Metis to accelerate agentic commerce

Summary

DoorDash has acquired Metis, a YC-backed applied AI research lab, to build agent-driven experiences that solve real problems for consumers and restaurants. Andy Fang, DoorDash co-founder and head of consumer engineering, said the company met with Metis around September 2025 and acquired them after working together on customer-facing AI workflows.

DoorDash is moving past the early approach of bolting chat features onto its app, which did not work well when LLMs were less capable. The core problem Fang identifies is decision paralysis. Customers tell him DoorDash has enormous selection but they have no idea what to order. Agentic commerce means using AI to help customers find what they want and help restaurants optimize pricing, promotions, and demand smoothing between dayparts.

Fang stresses positioning these features around utility rather than AI branding. "We wanna introduce it in a way that's actually beneficial," he says, avoiding the trap of slapping "DoorDash AI" on products just for novelty. The challenge is building experiences that feel natural and fast, not laggy or obviously algorithmic.

On the restaurant side, a small owner does not know how to price pad Thai in their neighborhood or what promotions to run. On the consumer side, one use case is automating lunch orders through an agent that checks your calendar, knows your preferences and favorite restaurants, monitors real-time inventory, and proactively places orders without you thinking about it. "If someone's building DoorDash today, it would not look like DoorDash," Fang says, noting that agents in an ideal world do not just inform but act on your behalf.

Metis brings AI research expertise and founder experience. The CEO is a former Mercor employee. Fang sees this as complementary to DoorDash's data advantage. DoorDash operates a large courier network and collects billions of data points in the physical world daily. Metis's experience in expert data helps DoorDash explore how to leverage that signal for robotics and physical intelligence, building on the dot delivery robots Stanley has been developing.

Fang mentions an OpenAI partnership where DoorDash integrated into ChatGPT, though the integration is limited. ChatGPT does not currently expose the partnership through Codex, the coding-focused CLI. He frames this as experimental territory. "A lot of people are just trying to throw stuff at the wall and see what sticks."

Fang notes interest from founders asking when a DoorDash CLI will launch. He is aware of vibe-coded Chrome extensions that filter DoorDash for specific dietary preferences but has not committed to a formalized developer ecosystem yet. He prefers to focus internally on core problems rather than encourage external engineers to rebuild DoorDash from scratch.

Fang also touches on Tempo, a stablecoin platform launching the same day, which DoorDash was an early design partner on. He hints that Tempo's work on stablecoins could be synergistic with agentic commerce, though he defers details to Tempo's founder, Matt.

All three DoorDash founders remain at the company 13 years after founding. Tony Xu stays in the CEO role. Stanley Tang and Fang focus on innovation and strategic bets without stepping on each other's toes. Fang credits clear role delineation and high trust.