How Walmart could have been early to the future of personal shopping agents

Good morning.

Another week, another big AI personal agent launch (or two).

OpenAI on Tuesday announced Dots, its twist on a personal agent that is rolling out first to paid ChatGPT users only.

“Your dot can take a project and run with it, even while it’s working on several others,” the company said in its announcement. “You can keep bringing it new tasks and ideas without having to juggle separate threads or direct every step.”

Friend of The Aisle Casey Newton said after a few hours of early testing, Dots is the personal agent he’s been “most impressed by.”

“It's highly capable, focused on the work I actually do, and intuitive to use,” he wrote.

Stratechery’s Ben Thompson said he was also intrigued by Dots, but did not see it as a direct competitor to Muse, at least for now. To him, it was more “chief of staff for everything you are working on” than a do-everything agent meant to help you run your personal life. I hope to spend some time with Dots later this week and will come back with my first impressions soon.

Elsewhere, former Yahoo CEO Marissa Mayer announced her own AI agent on Tuesday. It’s called Dazzle. The pitch? It’ll try to interpret problems or tasks captured in photos and screenshots in your camera roll, treat those images as prompts, and offer to take action on them, with little typing needed to get started.

Anything that attempts to make sense of my ridiculous number of forgotten screenshots on my iPhone is something I want to try. Who’s got an invite for me?

Then there’s Noah Shinn, the 23-year-old founder of the viral AI personal agent Instinct. Shinn stepped out of the shadows in a podcast interview this week, and he revealed some striking numbers to point to as evidence that agentic commerce is becoming real.

He said roughly 40% of Instinct users have handed the new AI personal agent a credit card thus far, and said the startup will eventually generate revenue from taking a cut of transactions it facilitates. He added that Instinct is already approaching $1 billion in annualized transaction volume, with about half of it from travel-related actions its service is handling on behalf of its users. No, he didn’t spell out the timeframe used to calculate that run rate. One day, one week, one month? Still a big number any way you slice it.

Finally, a piece of unfinished business from the RetailClub conference: Since I had to miss the event, I challenged attendees on LinkedIn to take the most fun photo in front of my very empty media tent. The winners are two Target execs, thanks to the “finger heart” that Target CIO Prat Vemana is flashing. Their prize? A coveted The Aisle-branded hat. They may have to fight over it, though I’m feeling generous enough that I might just send them one each.

I’ll have more insights and gossip from RetailClub over the next week.

Now on to the good stuff…

On an unseasonably warm winter day in March 2022, I sat across from then-Walmart CEO Doug McMillon in the same office that company founder Sam Walton once worked out of in the retailer’s original Bentonville “Home Office.”

I was there to interview him for my book, Winner Sells All, which would publish the following spring and chronicled the decades-long rivalry between the legendary retailer he was leading and the one-time pesky upstart called Amazon.

I only had 60 minutes with McMillon, but I guess he enjoyed the conversation enough that we ended up going about 90.

So I asked him about an internal strategy document called Walmart 2040, written by former Walmart e-commerce chief Marc Lore and former Walmart strategy executive Sloane Eddleston, which I reported on exclusively in the book.

For the prior decade, the online retail wars had largely been fought over three things: price, selection and convenience. But the memo’s authors believed Walmart would have a very difficult time beating Amazon on any of them.

Amazon had a vastly larger online assortment, while Prime and the company’s enormous fulfillment network made it extraordinarily difficult to surpass on convenience. The revenue generated from Prime memberships, online advertising, and Amazon seller fees also helped Amazon keep prices low.

So Lore and Eddleston proposed a different way to win: conversational commerce. Instead of asking customers to sift through an endless digital aisle, Walmart would create a personalized, one-on-one shopping experience at massive scale.

A customer could simply say what she wanted, or describe the person she needed a gift for, and Walmart, with the help of technology, would do what the company had done well since Sam Walton built it: Curate the best options for her. And not only that, then execute the purchase and delivery too.

The bet, in short, was that hyper-personalization could beat the unlimited selection of The Everything Store. What’s more, Walmart had already run an experiment, albeit an expensive one, that suggested the 2040 idea might have some merit. Customers of an internal Walmart startup called Jetblack ended up shopping on Amazon less frequently than they had before they used the upstart concierge service.

Now, four years after my conversation with McMillon, that entire vision suddenly looks a lot less futuristic; just last week, Meta announced that Walmart was among the retailers that would partner with the Muse personal agent.

What’s been overlooked in most of the coverage I’ve seen is that Walmart came remarkably close to giving its early experiment in a personal shopping agent a much bigger life. Lore had lined up a plan to spin the business out into a standalone company, with major outside backers ready to invest.

Walmart ultimately walked away.

But the details of how close that deal came, why McMillon killed it, and why Lore still thinks Walmart gave up something much more valuable than a money-losing concierge service are, I think, even more interesting now than they were years ago. So too are the thoughts of Jetblack’s former CEO, who has watched her vision come to life recently, with pride but also a twinge of envy.