DoorDash launched a new standalone 'Tasks' app on March 19, 2026, that pays its delivery couriers to complete assignments aimed at improving AI and robotic systems.
The tasks include recording unscripted conversations in Spanish, filming themselves completing household chores like loading dishwashers, handwashing dishes, or folding clothes, and holding objects up to body cameras for visual data capture.
The original audio and video footage is used to evaluate in-house AI models as well as those used by partners in retail, insurance, hospitality, and technology sectors. The goal is helping AI and robotics models better 'understand the physical world.'
DoorDash is making Tasks available to active couriers in select U.S. markets, notably skipping tightly regulated areas like California, New York City, Seattle, and Colorado. The company plans to expand into more task types and countries in the future.
With access to 8 million couriers who are already navigating the physical world daily, DoorDash is positioning itself as a data collection infrastructure layer for AI — turning its gig workforce into a distributed sensor network for training next-generation AI and robotics models.
The move highlights a broader trend: the gig economy is becoming the training layer for AI, with companies leveraging existing workforces to gather the real-world data that AI systems need to operate in physical environments.