The idea, which Thomas made a passing reference to during a keynote speech at CTIA Wireless, was referenced this way: "The best shopping list is the one you don't have to create and that's what we're working on." (Technically, the best shopping list is the one that someone else has to shop and pay for, but I digress.) Presumably, this mobile app would be built atop the chain's experimental Scan & Go mobile app, which prepares in-aisle checkout leveraging existing self-checkout units.
Given that Scan & Go—by its very nature—requires the shopper to register beforehand and to be associated with a verified payment method, it delivers an ideal CRM platform. This is a nice backdoor way to get into CRM for Walmart, which doesn't have a traditional CRM program and never has had one.
That is a crucial element of Thomas' self-populating shopping list. It would presumably build on that shopper's complete shopping history, noting typical durations between purchases of identical items and using that data to project when that shopper will likely be about to run out of that product.
The small magic here is that this is extreme customization, down to the individual shopper, and it's also likely—if done well, which Walmart has a nasty habit of doing—that it will be quite accurate. Combined, that delivers the big magic: namely, that this app is likely to be impressively sticky. In non-Web English, this is something shoppers are likely to use and to use often.
A good analogy for this is those automated reminders that car companies and gas stations like to send out via e-mail for oil refills. What should be a clever way to get customers back to pay for more service—presented as a courtesy FYI that your oil should need changing right about now—often fails. Personally, my driving patterns are erratic, and I rarely come anywhere close to driving the average number of miles per year. Therefore, those reminders are humorously way off. That means that I ignore them, which is the kiss of death for such functionality.
If Walmart's shopping lists are accurate, though, shoppers would have an excellent reason to routinely check them (while in-store), even if they think they have everything. Indeed, the auto-generated list could be better than a handmade list. Shoppers typically write down items that they've run out of and maybe where a family member has pointed out that something is about to run out. An app that uses historical buying from that person and projects likely imminent outages is powerful.
When the shopping history gets lengthy enough, the potential of this functionality gets even better. With a few months of data, it could make superb guesses as to when milk, eggs, bread or orange juice are likely to run out. With a few years of data, it could flag in July that the shopper almost always buys cranberry juice. (Turns out this shopper's cousin's family always spends a few days with them on the Fourth of July and one of their kids lives on cranberry juice.) It could anticipate and flag charcoal—or liquid propane—in June as well, if that's when that shopper typically stocks up.
It could also be smart about things like diapers and baby food, knowing which related items should be recommended. If that item suddenly pops up on the list, a checklist of related items might be quite well received by a new mother. The app could also be smart enough to know to stop recommending infant diapers after a certain amount of time and to change its baby-food recommendations to coincide with both age and changes in diaper size. It could flag any purchase of honey less than a year after diapers appeared, with a note that anyone younger than 12 months should not consume honey. (This could be boring or irrelevant to some but a literal life-saver for others.)
Thomas touched on the obvious, that it could also factor in dietary or budget restrictions. But to the extent that it would do that by examining the full ingredient list of every item (especially prepared foods, which tend to have a huge ingredient list), it could become indispensable.
That's why this specific bit of possible functionality has so much potential. For lots of reasons, this is one that shoppers are likely to actually use and use often. Then, when you layer discounts and special promotions atop, you have the ability to radically change buyer behaviors in-aisle.
If Walmart starts to control even a healthy minority percentage of every shopping list on a mobile in America (and beyond), a lot of retail execs are going to start losing a lot of sleep. Maybe Walmart will know to add warm milk, whisky and melatonin to their next shopping list?