Nicholas Upton
Artificial intelligence has finally proven itself as powerful as all the hypesters said: You can now rank all the restaurants in New York, San Francisco and Los Angeles by how hot the diners are.
It’s true. Thanks to AI-powered website LooksMapping, you can have a $100 billion data center tell you how hot, old or gendered a restaurant’s clientele is. Looking for hot young dudes? Check out Gurun Kitchen in San Francisco. Seeking the typical older woman? Check out Lennox Pollo in Los Angeles. Do you want something absolutely average? Join the middle-aged, passably attractive, gender-neutral haven Pio Pio 7 in New York.
Whether you think that’s brilliant or absolutely frivolous, it’s one more novel bit of AI that captures this moment well. Between the interesting bits and bytes, there seems to be a lot of quiet exploration of what might be the next internet, but with the sanguine understanding that—like the early web—a lot of it is rather useless.
Speaking at the Food On Demand Conference this spring, Leon Davoyan, chief technology officer of Dave’s Hot Chicken, said it’s time to test, iterate—and be ready to pull the emergency brake.
A voice AI pilot program worked from an order accuracy perspective, he said, but the technology harmed the cool, irreverent brand vibe. Davoyan said when guest sentiment scores began dropping, Dave’s “killed the project” right away.
“Even if a technology project is successful, if it’s at the cost of the guest, then it needs to be pulled,” he said.
Also in the voice AI arena, Taco Bell is seeing some progress and bumps. The brand rolled out AI to hundreds of drive-thru speaker boxes across the country in the last year, but is rethinking the use cases. Chief Digital and Technology Officer Dane Mathews told the Wall Street Journal some customers just don’t like it, while others mess with the AI to send insane orders to the kitchen, such as 18,000 cups of free water.
“We’re learning a lot, I’m going to be honest with you,” Mathews said in August. “I think like everybody, sometimes it lets me down, but sometimes it really surprises me.”
Softer AI steps
What does a quieter AI era look like? It looks like a bunch of smart folks hammering at the new tools, sans the fanfare of yesteryear. In September, McDonald’s hired Das Dasgupta, who has a long AI track record at McKinsey, Amazon and Starbucks, as its global chief data analytics and AI officer. Investors didn’t seem to care and the stock didn’t jiggle despite his success at creating easy and profitable AI wins at other mega corporations.
Similarly, technology providers are finding uses that aren’t grabbing big headlines. Checkmate, for example, rolled out AI-powered digital menu boards that transforms static screens to adaptive, intelligent signage.
“Digital menu boards have always been treated as static screens. At Checkmate, we see them as an e-commerce engine, a platform that continually experiments and optimizes to drive sales,” said Vishal Agarwal, Checkmate’s CEO, in a statement. “With AI, we can generate thousands of variations of banners and menu layouts within brand guidelines, then automatically test and self-optimize. This isn’t just a display; it’s a revenue channel built to adapt to every restaurant’s operations.”
The company worked with AI and machine-learning scientists to develop and test the technology. Unlike voice AI, consumers may never know they’re seeing one of multiple test screens or a menu designed to inspire them to buy a bigger afternoon snack during shoulder hours.
Starbucks showed the promise of AI helping with tedious tasks is still alive, and some of it is working. The company unveiled an AI-powered inventory system that cuts staff time from an hour a day down to 10 to 15 minutes. The tool, developed by Starbucks and vendor NomadGo, is essentially a tablet app that counts goods such as milk, syrups and cups. Staffers point their tablet into the fridge, it counts and records everything and they get on with their day. One can see the next step of using existing or new cameras to give this task entirely to AI.
There is also progress in some back-office categories, including marketing. We’ve all seen the churn of AI slop out there, but with some smart people behind the tools and guardrails on the robots, brands can get even closer to consumers.
As Jane McPherson, the senior vice president of marketing at Penn Station East Coast Subs, noted at FODC, she uses AI to tailor marketing messages based on behavior at an extremely granular level.
“We can really use generative AI at scale to customize the message to the consumer based on their purchase history, what they bought from us and when they purchased from us last,” she said.
New hype for agentic AI
Agentic AI is grabbing headlines and providing some exciting action. Walmart garnered attention and boosted outlooks from stock market analysts. The retail giant released its AI agent, Sparky, which helps shoppers find goods, sift through reviews and learn about products. The move has Walmart being hailed as a leader in agentic AI, and its stock approached record highs on the news.
Chipotle is similarly employing an AI agent named Ava Cado, which is reportedly helping the brand hire employees 75 percent faster. It asks and answers questions, and schedules interviews, taking a significant day-to-day administrative burden off store-level managers.
Chipotle Chief Human Resources Officer Ilene Eskenazi told CNN that candidates go from application to ready to hire within three and a half days, compared to 12 days with the traditional system.
Maybe the best thing about this new era of AI, with real data and some history? It’s getting easier to differentiate the real potential from the total nonsense.
Ivan Matkovic, CEO of loyalty provider Spendgo, stopped short of calling his fellow tech vendors liars at the recent FSTech conference, instead likening some of the hype purveyors to Tesla CEO Elon Musk.
“I think we’re in the assisted driving stage of AI, not self-driving,” Matkovic said. “A lot of vendors are overeager and have a lot of that Elon in them.”
Nicholas Upton has reported on retail and restaurant technology for more than a decade. His Tech Stack column aims to distill complex ideas into actionable insights. Send interesting tech topics to [email protected].