When we talk about the impact of AI on industries, we tend to think about how roles that are done by humans will be automated or ways that existing tasks can get easier. That’s important. But, AI, like other disruptive new technologies will enable significant changes in how we do things that are hard to anticipate. Imagine trying to explain to someone 30 years ago that they can pull a device out of their pocket and buy anything they want and have it arrive the next day. People would’ve found that hard to believe even though cell phones and the internet existed in early forms at this point.
I’ve been thinking about these unexpected changes that AI can bring to an industry. Today, I’ll talk about retail, and in future posts, I’ll talk about ideas for other sectors.
One very exciting AI technology that will have a huge impact on the world is self-driving cars, and self-driving cars are now real. If you go to San Francisco or Phoenix, you can summon one as easily as ordering an Uber. This has implications for all kinds of parts of the economy: long haul trucking, taxis, elder care, commuting, etc. But one rarely discussed place where there could be some big and exciting changes is retail.
As mentioned above, the biggest disruption in retail in the last 30 years is e-commerce. It used to be that if you wanted a pair of jeans, you had to drive to a store, try some on, and then buy them. Now, many people go online, order a bunch of pairs, including some in various sizes. Then, they try them on and return most of them. The retailer loses on shipping fees and may not be able to efficiently restock the items with some ending up being written off.
But what if there was a third way to buy stuff? A new method where the store comes to you. Well, it might be possible soon…

The Concept: Self-Driving Retail Vehicles
Imagine a fleet of specialized, self-driving vehicles, designed to cater to different retail needs. Whether it's clothing, electronics, or beauty these autonomous stores on wheels could revolutionize the way we shop.
Consider the "jeansmobile" as an example. When you need a new pair of jeans, you simply use an app to summon the nearest jeansmobile. Within minutes, the vehicle arrives at your location, stocked with a variety of jeans in different sizes and styles. You step inside, try on the jeans in a private, comfortable space, and make your selection. There’s no human driver or clerk, but there’s a virtual assistant who is happy to weigh in on whether the jeans make your butt look good. The vehicle then processes your payment and drives off, leaving you with instant gratification. You have the perfect pair of jeans in your hand less than an hour after you ordered them with no waste! (Although unless the car is electric and charged by green energy, there is some environmental impact.)
Could the Economics Work?
We are at very early days on self-driving cars, but I penciled out some rough numbers, and it seems like it could probably work1:
Revenue assumptions:
The van can serve 12 customers a day (note that it could run 24 hours, so this is conservative)
Average order is $100
300 days a year in operation
Revenue total: $360,000 per year
Cost assumptions:
COGS = 50% of revenue = $180K
Fixed costs: $60K
Depreciation of van over 5 years (@$200K) = $40K
Insurance = $10K
Tech = $10K
Semi-fixed: $44K
Maintenance = $4K
Energy = $6K
Staffing to manage business = $22K
Marketing = $12K
Annual profit = $76K per year, so payback period on the $200K van is 2.8 years.
Presumably, if you tried to scale this up, you could get meaningful economies of scale, etc. Also, note that if this was not self-driving, the math gets much harder. Suppose a driver/clerk costs $50K a year all-in, and the new van with a driver costs $100K (instead of $200K), and sales per day are 10 (vs. 12 because the person works shorter hours). That gets us to $300K of revenue, so -$60K which is -$30K of gross profit (still 50% gross margin). Costs go up by net $30K ($50K people cost - $20K lower depreciation). So, net impact is -$60K which means profit is $16K per year with a human or a payback of 6.2 years, outside the 5 year lifetime of the van.
I’m sure you could debate these assumptions and arrive at different answers, but the point is that this feels like a viable business idea with self-driving cars that wouldn’t be without them. It especially works in a densely populated area with high demand.
What Else is Possible?
If you accept the idea of on-demand mobile retail, then this idea could work in lots of other domains:
Other apparel categories where it’s hard to know fit: Think sneakers, dress shoes, kids’ clothes. Same idea as jeans – you try them on and see what fits and looks good before you buy them.
Electronics and gadgets: Need a new laptop or phone? A techmobile stocked with the latest gadgets could come to you, allowing you to test and compare products before making a major purchase. You could feel how heavy the device is, see how big the screen is, see how sensitive the touch screen is, etc.
Beauty and personal care: From skincare to makeup, a beauty mobile could offer personalized consultations and allow you to try products before buying.
The Road Ahead
Autonomous vehicles are offered now in targeted markets, but we are a long way from widespread adoption of the technology and full regulatory approval. It will be a couple of years before firms could license this technology to make this service a reality. Yet, I wouldn’t bet against on-demand retail being how we buy clothing in 2030.
If you are an investor or executive at a retailer, maybe it’s time to start piloting this now with a human driver to get ahead of the curve.
Sound off in the comments if you think you’d actually use this service. I think I would.
I'd give this a go and would be excited about this as a way to shop assuming there is sufficient choice! Very interesting business model. I imagine one of the NB things to figure out is how to estimate the propensity to transact for any customer that is requesting the vehicle. I guess you could also charge a base fee when the initial request is made to serve as a deposit and encouraging the customer to transact.