Starbucks recently launched a new ordering feature integrated with ChatGPT, aiming to simplify the coffee purchasing process through conversational AI. However, user testing of the beta integration revealed significant usability issues that rendered the experience frustrating rather than efficient. The core promise of the feature was that users could simply type their order, such as "Order me a Venti iced coffee with light skim milk," into the ChatGPT interface and have it automatically added to their cart for pickup. In practice, the system failed to execute this streamlined transaction. Upon receiving an order request, the AI did not immediately place the item. Instead, it provided descriptive feedback about the drink and presented a menu of likely options. Users were forced to manually navigate a complex interface, selecting customization options like size and milk type from pop-up UI elements. Without these additional clicks, the system would default to a different specification, such as a Grande black iced coffee, defeating the purpose of the natural language command. This multi-step process proved slower than the traditional Starbucks mobile app, which typically requires only four taps to replicate a familiar order. Further complications arose when attempting to add multiple items to a single order. The AI struggled with vague beverage names like "the fruity tea," often suggesting incorrect alternatives like Iced Green Tea Lemonade instead of the correct Passion Tango Tea. Additionally, users encountered strict usage limits on the free tier of ChatGPT. A pop-up warning indicated the chat session was nearing its limit, eventually resetting the user to a less capable model that could no longer access advanced Starbucks features or place orders directly. Technical glitches compounded these issues, with the system failing to detect the user's location correctly and offering an error message when attempting to update the store selection via a map view. The fundamental flaw of the integration lies in its reliance on a conversational format for a task that is inherently transactional. Coffee ordering is a routine action where speed and accuracy are paramount. The chat-based approach introduced unnecessary back-and-forth interactions that offered no discernible value over standard menu navigation. While Starbucks' promotional materials suggested creative use cases, such as recommending drinks based on a user's outfit or mood, these features do not address the practical needs of daily customers. The experiment highlights a recurring challenge in the tech industry: attempts to apply generative AI to simple, repetitive transactions often result in a degraded user experience. The true potential for AI in this sector may lie in autonomous agents that can navigate applications to complete tasks silently, rather than chat interfaces that require users to micromanage the process. For now, the feature appears better suited as a novelty than a functional tool for the average consumer.