Key Takeaways

  • DoorDash has been building in robotics since 2018, long before the wider hype, showcasing a deep, long-term commitment.
  • Traditional autonomous vehicles (like slow sidewalk robots or large robo-taxis) didn't fit DoorDash's specific need for 3-5 mile deliveries in dense suburbs.
  • This led them to a "first principles" approach, designing DOT: a 300 lb robot capable of 20-25 mph, modeled after an autonomous scooter, not a car.
  • The key insight: goods delivery faces a unique “first and last 100 feet problem” that passenger transport doesn't, demanding a purpose-built solution.

The Method

Stanley Tang explains DoorDash's journey into autonomous delivery by highlighting a crucial mistake many builders make: starting with the tech, not the specific problem. “It's people just kind of do the opposite where they try to build the tech first and and and not really think about the use case they're building towards,” Tang said. This common pitfall, he explains, results in expensive, oversized solutions that simply don't fit. DoorDash, however, chose a different path after years of partnerships and exploration, "since 2018 actually," Tang noted, reflecting their long-term, patient approach.

Their method for creating DOT was the opposite. First, they deeply understood their specific use case: “that 3 to 5 mile delivery in dense suburb, which is where most of the deliveries happen.” They ran experiments, looked at their "10 billion delivery" data, and partnered with various autonomous vehicle companies, but nothing truly matched this specific need. Sidewalk robots were too slow for a 3-5 mile trip, unable to deliver efficiently across those distances. Bulkier robo-taxis, designed for people, were completely "different" for goods. “If you only have if you're only carrying a couple burritos around, do you really need a 4,000lb car with chairs and AC?” Tang asked, pointing out the absurdity of using an oversized solution. The core problem, he argued, around “carrying people and carrying goods is actually a little bit different.”

This realization led them to a "first principles" redesign. Instead of adapting existing tech, they imagined the ideal delivery vehicle for their precise constraints. Tang describes the right "metaphor" as “probably an autonomous motorcycle or scooter or bike profile vehicle.” Their new robot, DOT, weighs around 300 pounds and can travel at 20-25 mph, directly addressing the speed and size requirements for efficient suburban delivery. This bespoke design also specifically tackled “what I call the first and last 100 feet problem,” the unique challenge of getting goods from the vehicle to the customer's door, which differs greatly from passenger drop-offs. “We're going to control our own destiny here,” Tang explained, reflecting their decision to invest in building in-house after realizing existing solutions were a poor fit.

Where This Breaks Down

While powerful, DoorDash's first principles approach isn't always the right path. This method demands significant upfront investment in time and capital. DoorDash started exploring robotics in 2018 and spent years on partnerships before committing to building DOT in-house. Most early-stage founders won't have the runway or resources to “control their own destiny” on this scale, especially if a "good enough" off-the-shelf solution could get them to market faster. This approach also requires deep, proprietary insights into the problem, often derived from vast amounts of data like DoorDash's "10 billion delivery" history. Without such granular understanding, you risk building an expensive custom solution that's no better than existing options.

What to Do With This

As a founder in your 20s or 30s, identify a core operational pain point or customer interaction where existing tools or "best practices" feel slightly off. For instance, if you're using a generic CRM for a sales process that has extremely niche compliance steps, or if your onboarding flow for specific users involves workarounds. Map out the precise constraints of that problem, similar to how DoorDash defined their 3-5 mile suburban deliveries. Then, ask: "If I built the absolute minimum viable thing to solve just this problem, what would it look like?" Design that ideal, lean solution from scratch, even if it's just a custom internal script or a simplified user interface, rather than forcing a 4,000lb software suite onto a burrito-sized problem.