Why We Invested in Kiwi Campus: The Future of Food Delivery
We’re thrilled to announce our partnership with Kiwi, an ambitious team focused on dropping food delivery costs through sidewalk delivery drones. Welcome Felipe and the Kiwi Team to the Prototype Capital family!
There are some of the fundamental truths we came in with:
- Delivery will be an increasingly important part of food and people have high expectations: The overall shift to a take-out dominant food service industry is inevitable as consumers began to rely on convenience. McKinsey reports consumer’s expectancy in food delivery has seen a significant shift from satisfactory at a 60 minute arrival time to a 15 minute arrival time. Restaurants are now noticing the shift in consumer tendencies and are decreasing dine-in amenities to open up margins for delivery infrastructure (including secondary online ordering food production lines) which now amounts for up to 50% of sales at brands like Sweetgreen.
- Long term cost of delivery will asymptotically approach zero: As consumers begin to expect low delivery times, the only way companies can open up to more consumers is by lowering prices under the traditional $8 fee, a typical fee today. Just as Uber and Lyft, consumers for delivery will be relatively brand agnostic and price inelastic; thus for any arbitrary market, whichever company can have the lowest fees (for the same delivery time) will win that market. And the market is a winner take all market just as today certain markets are dominated by either Caviar or Postmates or Uber Eats or Doordash.
- Robots are perhaps the only way this will happen: George’s Institute of Technology finds that there are two primary problems around a courier-based delivery system: courier assignment (dynamic vehicle routing) and capacity management (offline shift scheduling). In exchange for internalizing some of the risks and costs associated with demand uncertainty, couriers are entitled to a significant degree of autonomy, thereby adding yet another layer of complexity. Whereas robots will implement deterministic instructions from central planning, humans as independent contractors might do so; for the latter, this introduces uncertainty in scheduling (couriers have some freedom to choose the hours they work), dispatching (couriers can reject an offered assignment), and routing (couriers can disregard the suggested sequence of deliveries); this all adds extra unit costs and people overhead, leading to a less sustainable long term business than robotic couriers.
- For autonomous vehicles, it’s better to be making deliveries with a “human in the loop” before launching fully autonomous vehicles: In the race to a autonomous delivery system, companies like Nuro and Starship are building fully-autonomous systems complete with the best parts and with LiDAR. We believe a superior strategy is to achieve partial autonomy in conjunction with manned remote centers that supervise bots on the streets in case of algorithmic perception mistakes. This would allow the company to collect mass amounts of training data and train consumer behavior while it builds out it’s fully autonomous system (as opposed to launching all at once the company has achieved full autonomy).
- For the unit economics to make sense, the cost of the robot has to be quite low: This is the reason we believe delivery bots on the sidewalk will beat out cars (a la Uber Eats) in the long run. When you have a system that’s $40k (a car or a fully-autonomous highly-customized bot with LiDAR, say), you need to do significantly more orders to return the cost of capital of hardware. On the other hand, if you take a purely computer vision approach to autonomy using stereo cameras (a la Tesla) you can significantly improve your unit economics to build a more sustainable business (that’s not completely dependent on venture dollars). The lower the cost of the bot (without of course sacrificing quality), the better.
- It’s better to work with the government than to fight it: As the Bird incident in San Francisco has shown us, rather than begging for permission (the Uber / Airbnb style), today it’s perhaps better to ask for permission upfront.
CEO and co-founder Felipe besides the first Kiwi Bot.
We believe Kiwi offers the best solutions to these trends and realizes these industry truths. The team at Kiwi has not only launched a beautiful mobile-facing point of service to order, but with over 23,000 deliveries already over the past two years, Kiwi has taken the city of Berkeley by storm (their competitors who have raised 50x the money they have have done zero or next-to-zero). The Kiwi bots and team leverage a hybrid of robot/human interactions to decrease delivery times, delivery costs, and the complete end user experience. The Kiwi bots have brought the people of Berkeley, CA a full stomach and a caring community around the lovable, emotion-emitting automated couriers.
Kiwi is already lowering food delivery costs in their market. Kiwi’s logistics model allow for current delivery fees to profitably stand at $3.20 (and falling) compared to competitors at $10 (which even then are often subsidized via venture dollars).
This industry is no doubt hard, combining food and perishables with delivery and logistics with hardware with autonomous vehicles with a highly demanding consumer with government regulation… Ultimately, our decision to work with Kiwi was mainly driven by the team behind the bot. Felipe’s unchallenged competitive slant and raw relentlessness to break through brick walls in order to cost of delivery for a take-out dominant future makes him the ideal general in this war of attrition for low-cost food delivery. Kiwi’s business and product is loved among the Berkeley community with students who care for their tiny stewards, advocate for their safety (a crowd of passerbyers on campus shrieked when someone was about to kick a bot for fun), and even dress as Kiwi bots for Halloween. We’re thrilled to be apart of this journey alongside Felipe and the team at Kiwi in dropping the time and cost of ordering any spectrum of foods.
To a beautiful future,
Team Prototype Capital