Starbucks and Little Caesars launched apps inside ChatGPT this month. Both brands are betting consumers will order food the same way they already plan meals and research products. The interface is familiar. The checkout is not.

The Setup Takes Longer Than the Order

Getting started requires work. Users navigate to a connector directory inside ChatGPT, search for the brand and enable the integration manually. None of that happens in the native app. The Starbucks app opens with saved preferences, a loyalty balance and a reorder button. The ChatGPT path opens with a blank prompt.

PYMNTS asked for something low-fat and high-protein. Starbucks returned a relevant drink, explained the option and let size selection happen inside the chat. Nearby locations appeared next. The flow worked. Then a store got selected. A browser opened. The Starbucks app took over. Login, payment, checkout. The conversation did not carry over.

Little Caesars did not get that far. The same ask returned recipe suggestions and outside options. Nothing came from the actual menu. A follow-up prompt produced one item. The flow then asked for a ZIP code. No results came back. Checkout never happened.

Where the Handoff Breaks

Both brands hand off to an external screen to complete payment. Starbucks routes checkout through its own app or website. Discovery happens inside ChatGPT. Payment happens outside it. The company made that choice deliberately as its loyalty program depends on owning the transaction.

That split reintroduces the friction the channel was designed to eliminate. The native Starbucks app completes an order in second, holds users’ saved payment credentials and loyalty state. The ChatGPT flow takes several minutes to reach the same point. The handoff to a login screen does not feel like a final step; it feels like starting over.

The back-end infrastructure is the problem. Payment credentials live in one system. Loyalty lives in another. Order history lives somewhere else. The conversational front end cannot connect them. Users bridge that gap themselves, adding steps instead of removing them.

What the Friction Is Actually Measuring

This is not a technology failure. Commerce infrastructure is not built for this.

Bites is building its entire business model around ChatGPT as an ordering channel. Starbucks kept checkout inside its own ecosystem on purpose. Neither brand is treating this as a gimmick. Both are making deliberate bets on where the interface is going.

Those bets have logic behind them. Conversational ordering removes the browse-and-scroll loop. A model that knows a user’s preferences, location and time of day can outperform a static menu. The discovery layer already shows what that looks like.

The checkout layer is not there yet. Credentials need to travel with the conversation. Loyalty state needs to persist across sessions. Payment needs to complete inside the chat. Until that infrastructure exists, users have to bridge the gap themselves.

The tools work. The plumbing does not. That gap will close. The question is whether consumers wait for it or return to the three-tap experience that already knows their order, their address and their card number. Habit is a short-term moat. So is convenience. Right now, the native app still offers both.