Until this week, the only way to send one of Serve Robotics' sidewalk robots down a street was to be Serve, or one of its handful of restaurant partners. On November 11, that changed: the company opened an AV Delivery Developer Portal, a public API that lets outside developers, businesses, hobbyists and AI agents programmatically dispatch its fleet, according to The Robot Report.
The fleet behind the API is already sizable: more than 2,000 robots operating across roughly 150 neighborhoods in 20 U.S. cities, plus a presence in Australia, reaching an estimated 3 million people and serving more than 4,000 restaurants. Serve says the robots have completed hundreds of thousands of deliveries to date. The new portal includes a sandbox environment where developers can build and test integrations before touching a live robot, with access to real fleet dispatch requiring Serve's approval.
Turning a delivery company into a platform
"You shouldn't have to start a robotics company to build something useful with robots," said Ali Kashani, Serve's co-founder and chief executive, framing the move as an attempt to make sidewalk delivery into shared infrastructure rather than a service only Serve itself can sell. That positions the company less like a delivery operator and more like a utility other businesses plug into, similar to how ride-hailing and mapping APIs let outside apps build on top of existing driver and map networks.
The pitch specifically name-checks AI assistants as potential users, not just human developers, suggesting Serve expects demand to come from software agents booking deliveries autonomously on behalf of customers, rather than only from app developers writing traditional integrations. If that materializes, Serve's robots could end up fulfilling orders placed by other companies' chatbots and agents without Serve ever touching the customer relationship.
Why the timing matters for sidewalk robotics
Serve's move comes as the sidewalk delivery robot category consolidates around a handful of players operating at real scale rather than pilot programs. Competitors including Starship Technologies and Coco have built comparable fleets, but Serve's decision to open dispatch to third parties is a distinct strategic bet: rather than competing purely on delivery volume, it is trying to become the default dispatch layer other companies build on, the way cloud infrastructure providers compete on who else runs on top of them.
The approach also gives Serve a path to grow utilization without having to sign every new restaurant or retailer itself. A smaller business or app that cannot justify building delivery infrastructure from scratch can instead call Serve's API, potentially filling gaps in the robots' schedules between existing partner orders. Whether that demand shows up quickly, or whether approval requirements for live dispatch access slow adoption, will be the first test of whether treating a delivery fleet as a developer platform actually expands Serve's business beyond the restaurants it already serves.