A Pizza Hut franchisee is suing for $100 million after an AI rollout went wrong. The technology did its job. That is exactly the problem.
A Pizza Hut franchisee is suing for $100 million, and the headline everyone will reach for is "AI gone wrong." That is not quite what happened.
According to the claim, the AI did not fail. It worked. It did the job it was built to do. The business broke anyway.
Chaac Pizza Northeast runs more than 110 Pizza Huts across the US east coast. It alleges that a new AI dispatch system, rolled out alongside a national DoorDash deal, gave delivery drivers far more visibility into kitchen timing and order details than they had before. The drivers used that visibility. They waited to batch orders together, so pizzas sat out of the oven. Because they could see tip amounts, they picked the good orders and left the rest. Delivery times slipped. Customer satisfaction fell. Sales followed.
These are allegations, not proven facts. But the pattern in them is worth sitting with, because it is one I see again and again. The detail that matters most: before the new system, Chaac was one of the chain's best performers. Its older, more manual process delivered more than 90% of orders within thirty minutes.
Here is what the case has to teach.
1. Automation scales the process you already have
If the underlying process is sound, automation makes it faster. If it has gaps, automation makes the gaps faster too. Chaac's manual setup had quiet controls built into it. The new system stripped them out. Understand and fix the process before you put speed behind it.
2. A new tool changes who can see what, and people act on what they can see
The drivers were not malfunctioning. They were behaving rationally inside a new information environment. The moment they could see timing and tips, they had power they did not have before, and they used it for themselves. Before you switch a system on, ask who gains visibility, and how their behaviour will change once they have it.
3. Faster is not the same as better
The new system was more sophisticated than the one it replaced. It still produced a worse result. Speed and intelligence are not the goal. The outcome is. If a simpler process gets the pizza there hot and on time, that process is winning, however unglamorous it looks.
4. Buying a tool is not the same as adopting it
The claim says operators were not properly trained and requests for support went unanswered. A capable system landed on people who were not set up to run it, and the value leaked away. Most of the spend goes into switching the tool on. Most of the value depends on what happens after.
5. The second-order effects are where the damage lives
Nobody set out to slow deliveries. The harm came from a chain of small, predictable reactions that nobody mapped before launch. The failures that hurt most are rarely the tool breaking. They are the things the tool quietly sets in motion.
The order matters
None of this is an argument against AI. It is an argument for sequence. Sort the process first, understand how people and incentives actually work, then automate what is ready. Do it in that order and the technology earns its keep. Do it in reverse and you can spend a fortune making a good business worse.
At N16, that is the order we work in. Every time.
