Keep pulling the thread on Andy Fang and Stanley Tang.
50% of user journeys using the "Ask DoorDash" feature for restaurants result in customers ordering from a place they have never tried before.
Grocery orders made through the "Ask DoorDash" feature have basket sizes that are 40% larger on average.
DoorDash determined the ideal autonomous vehicle for its use case is a 300-pound machine capable of 20-25 mph, similar to a scooter or bike, to handle 3-5 mile deliveries in dense suburbs.
DoorDash completes over 3 billion deliveries a year, with no two deliveries being identical.
DoorDash has a dataset of 10 billion completed deliveries, which it views as a significant competitive advantage for developing AI and autonomy.
The DoorDash Dot is an in-house autonomous delivery robot that weighs 300 pounds, travels up to 20 mph, and is designed to operate on sidewalks, bike lanes, and roads.
DoorDash has been conducting live, fully autonomous L4 deliveries with its Dot robot in Phoenix for over a year.
DoorDash has a unique data advantage for solving the "last 100 feet" delivery problem by using historical drop-off locations from human Dashers, which is more precise than GPS data from services like Google Maps.
DoorDash's AI compute spend in June 2024 was 20 times higher than its spend in January 2024, but has since started to flatten.
DoorDash will have more human Dashers in 10 years, not fewer, even with widespread adoption of robotics and AI.
DoorDash's initial attempts at conversational commerce two years ago were focused on voice as the primary modality, which ultimately did not succeed.
The "Ask DoorDash" feature incorporates external "world knowledge," such as restaurants trending on the internet or in forums, to improve recommendations.