Here is one way the AI data economy works in practice in 2026: a DoorDash courier straps on a body camera, washes at least five dishes, holds each one up to the lens for a few seconds, and earns a few dollars. That footage, mundane, specific, reproducible at scale, is exactly what AI and robotics companies need to train models that understand physical tasks.
And it turns out that a delivery network of eight million people, already dispersed across almost every postcode in the United States, is a remarkably efficient way to collect it.
DoorDash on Thursday launched Tasks, a new product that formalises what had been emerging piecemeal across the platform for the past year. It operates on two levels.
The first is a set of new task types inside the existing Dasher app: taking photos of restaurant dishes to populate a menu, photographing a hotel entrance so future drivers can find the drop-off point, or scanning supermarket shelves for inventory checks.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
The second is a standalone Tasks app, designed for activities with no delivery component at all, filming household chores, recording unscripted conversations in another language, or, in a partnership that drew attention back in February, closing open doors on Waymo’s self-driving cars in Atlanta.
The Waymo door-closing programme, which DoorDash and Waymo confirmed to TechCrunch and Bloomberg in February, already sits inside the Tasks platform.
When a Waymo passenger leaves a vehicle door ajar, a safety trigger that prevents the car from moving, nearby Dashers receive a notification and can earn around $11 to drive over and close it. It is a small transaction with an outsized symbolic weight: gig workers, often cited as the group most exposed to displacement by automation, being paid by an autonomous vehicle company to solve a problem its own technology cannot yet handle. Waymo has said future vehicles will include automated door closures.
For DoorDash, the logic of Tasks is straightforward. The company has spent more than a decade building the operational infrastructure to dispatch workers to specific physical locations, verify completion, and handle payments at scale.
That is exactly the capability that AI data collection requires, and it is not something that can be replicated quickly by companies that do not already have a network like it.
“There are more than 8 million Dashers who can reach almost anywhere in the U.S. and who want to earn flexibly beyond delivery. That’s a powerful capability to digitize the physical world,” said Ethan Beatty, General Manager of DoorDash Tasks, in a statement.
The scale claim is significant: companies like Scale AI built entire businesses around remote data labelling workforces, and DoorDash is arriving in that market with a distribution network already in place, operating in-person rather than online, and capable of collecting the kind of embodied, physical-world data that is increasingly scarce and valuable.
DoorDash says Dashers have completed more than two million tasks since 2024, a figure that covers the earlier, lower-profile incarnation of the programme before Thursday’s formal launch.
The company is not the only delivery platform to have moved in this direction: Uber and Instacart have both introduced similar programmes over the past year.
There are questions the launch does not answer. DoorDash has not published detail on how it handles consent, data retention, or the rights workers have over footage of themselves in their own homes. The exclusion of California, New York City, Seattle, and Colorado, jurisdictions with significantly stricter gig worker and data privacy regulation than the rest of the country, is conspicuous.
Pay is determined upfront on a per-task basis, weighted for effort and complexity, but no average rates or floor guarantees have been disclosed. For a programme that requires workers to bring cameras into their kitchens and record their own voices, those are not minor details.
DoorDash says it plans to expand into more task types and countries. For now, what it has launched is a version of something the technology industry has been trying to build for years: a human sensing layer over the physical world, paid by the task, available on demand.
