Starbucks has a solution for indecisive coffee lovers faced with a truly dizzying array of options: simply let the software choose for you.
The Seattle-based chain is testing a U.S. beta app on ChatGPT that helps users discover drinks based solely on feel and intuition, the company announced — yet another step by big American companies to make consumer-facing activities less like browsing a menu.
Also read:
Continues after advertising
The beta app is built directly on ChatGPT, where users can enter a command describing their caffeinated needs and desires. Asking for “something sweet and nutty,” for example, might prompt a recommendation for a pistachio latte. Users can also upload photos and ask which drinks would aesthetically match their outfits or surroundings.
Starbucks joins retail and travel booking services that have already launched similar features offering personalized recommendations. Companies seek to gradually reduce the burden of choices in shopping, travel and food, but by easing the decision-making process for customers, they may face unexpected risks.
Take Walmart as an example. Last year, the company announced a partnership with OpenAI that would allow some customers to purchase products from Walmart and checkout directly through ChatGPT. The program started with around 200,000 products, and executives looked primarily at conversion rates — the percentage of consumers who actually complete a purchase.
The test was a failure. Conversion for products suggested by ChatGPT was three times lower for items that needed to be purchased within the app than for products that required clicking through to Walmart’s main website, according to a Wired report last month.
The company changed its strategy and started incorporating its own AI chatbot into ChatGPT to streamline processes, but consumers still seem to value the decision-making process that accompanies purchases.
Daniel Danker, executive responsible for AI products at Walmart, told Wired that one of the limitations of the experiment was that customers tend to decide to purchase accessories alongside the main product.
Continues after advertising
Discovery or repetition
The Starbucks app will be different because users won’t be able to make purchases directly through ChatGPT, but it could still change the way consumers find their next favorite item — possibly for the worse.
Some research suggests that AI recommendations could even take some of the excitement out of exploring products. A 2025 study from a Chinese university, published in the journal Advances in Consumer Research, found that AI-generated suggestions on e-commerce platforms tend to provoke more negative feelings among consumers.
While some value greater simplicity, many report feeling trapped in “information cocoons” in which AI assistants offer repetitive recommendations that reinforce existing preferences rather than directing toward interesting or unexpected products.
Continues after advertising
Starbucks representatives say their app can avoid this pitfall by reacting to described feelings or environmental cues rather than relying on the user’s history.
“We want to meet customers at exactly that moment of inspiration and make it easier than ever to discover a drink that matches,” said Paul Riedel, senior vice president at Starbucks, in a statement to Fortune. “This is the kind of technology that sparks creativity and helps customers discover something new.”
Companies like Starbucks that are experimenting with AI recommendations are doing so at a time when Americans are beginning to push back against interacting with technology in all aspects of life.
Continues after advertising
From resistance to use at work to rejection of AI-generated digital content, consumers are beginning to show fatigue — and even irritation — with technology. Brands are paying attention: some have already removed AI-created ads following online criticism.
A Harris Poll survey released this week showed that only 39% of American consumers trust AI to make everyday purchases for them, while the vast majority still distrust the quality of products presented online.
Starbucks itself has also made efforts to distance itself from the automation narrative and preserve in-person interactions with customers. Last year, CEO Brian Niccol announced that more stores would increase the number of baristas, with fewer automated options, and advocated for more personal touches such as “handwritten notes” accompanying orders and the use of ceramic rather than disposable cups.
Continues after advertising
It’s true that Starbucks customers would probably benefit from some light guidance when ordering. Between customizable sizes, number of espresso shots, types of sweeteners and all the extras the chain is known for, a simple latte order can unfold into more than 300 billion different combinations.
This huge list of possibilities has already been cited as one of the reasons for long waiting times at peak times. After taking office in 2024, one of Niccol’s first measures was to reduce the size of the menu by 30%, classifying many options as “excessively complex”.
AI recommendations can make customers’ lives easier when faced with so many options, whether searching for a specific model of sweater or a drink with the exact shade of caffeine. Companies try to turn this decision fatigue into a competitive advantage, but if customers are led to always order the same caramel latte combination, they may miss out on the pleasant surprises of freer exploration.
2026 Fortune Media IP Limited
