TL;DR: I spent a full week at FSTEC in Dallas, and AI was the first thing nearly every booth led with. Not one operator asked me about an AI strategy. They asked about Friday-night no-shows, a GM losing four hours a week to the schedule, drive-thru timers that get gamed, and delivery orders keyed in wrong. My rule for your next vendor meeting: nobody gets to say “AI.” Make them name the business problem, then ask three questions before you sign anything.
What is this episode about?
This is a solo RTG Insights episode, about twelve minutes, recorded right after a week at FSTEC in Dallas. It is about how operators should evaluate AI pitches: start from the business problem, not the technology, and judge every vendor on what changes for your guests, your sales, your profit and your team.
I have been in restaurant technology for more than thirty years, and I host the Restaurant Technology Guys podcast — more than 300 episodes since 2012, sitting down every week with operators and technology leaders about how they solve real business challenges. The pattern I see on trade show floors has not changed in three decades. Only the buzzword has.
Why does every restaurant tech vendor lead with AI?
Because it is the word that gets attention right now. At FSTEC, AI was what almost every booth led with, and it was the topic of many of the breakout sessions. But in a full week of conversations, not one operator walked up to me and said, “Tell me more about your AI strategy.”
Here is what they did ask me to help solve:
- How do I deal with guests who reserve a Friday-night table and never show up?
- How do I get my GM back the four hours a week they spend building a schedule?
- How do I fix drive-thru times when people game the timer?
- How do I stop third-party delivery orders from being entered into the system wrong, so the guest actually gets what they ordered?
Every one of those is a real business problem. None of them is “I need more AI.”
Is AI a strategy for restaurants?
No. AI is a tool, no different from your POS, an integration partner, or the knife, oven and grill in your kitchen. It is not a strategy, it is not a feature, and it is not a reason to buy something if you do not know you have the problem it solves.
I say it on the show all the time: technology for technology’s sake is worthless, but technology that solves a business problem is huge. Operators are not looking for AI. They are looking to get a job done.
What should I ban from my next vendor meeting?
The word AI. For your next meeting with a vendor partner, tell them they are not allowed to say it — and neither are you. Make them explain what their solution solves from a business perspective. If they cannot describe the problem without the buzzword, they probably do not understand the problem.
Do not let anyone talk you into the idea that AI is going to make everything go away. If it is not solving a real problem, none of it matters.
Haven’t we seen this hype cycle before?
Many times. I have watched operators get excited about the latest technology for thirty years — the internet, mobile ordering, everything moving to the cloud, third-party delivery, online ordering, tablets at the table. Each wave came with the same mistake: nobody stopped to ask what business challenge they were trying to solve.
Are we trying to increase throughput? Create better guest engagement? Fix something our managers deal with every shift? Vendors will happily tell you their solution makes all your problems disappear. In reality, every one of these projects takes change management and buy-in from the operators who live with it.
Consultants and vendors who work across a wide spectrum of brands tell me the same thing I see: people throw technology at a problem without really understanding the problem. I covered the broader version of this in 5 Crucial Tech Lessons Multi-Unit Operators Wish They Knew Before Scaling.
What happens when an operator buys AI on the pitch?
It usually shows up about six months later. The operator who buys because the demo was exciting, without being clear on what they were solving for, ends up with a problem that is still there, a pile of money gone, and a lot of time and effort spent getting nowhere.
That last part is the expensive one. Time and energy inside restaurant brands are at an all-time low. Managers are stretched, support teams are thin, and every new system competes for the same few hours. A tool that does not solve a real problem does not just fail to help. It takes hours away from the things that would have.
So before the pitch, clearly define what you are trying to solve, so you go into the vendor conversation asking the right questions instead of reacting to the vendor’s.
What is the Tuesday test for restaurant technology?
It is the test I would put every solution through. Not the Friday night. Not the day you got the demo. A standard, boring, slow Tuesday. What does this technology change on that day?
If it does not change something for your guest, your top-line sales, your bottom-line profit, or your staff, it is not a solution. It is software out chasing a customer who does not understand what they are trying to do.
That matters more with AI than with most waves before it, because many of the new AI players do not have any legacy in the restaurant space. Someone hits a problem, has a computer science degree, decides AI will fix it, and builds a product without fully understanding what a restaurant actually needs.
What business problems can AI actually help a restaurant solve?
Take AI out of the sentence and ask what you are trying to build. Is it:
- A better sales forecast for your manager?
- A better schedule?
- Better forecasted ordering?
- Enough staff in the kitchen to hit the rush, without overstaffing the slow periods?
- Front-of-house staff increasing sales and their tip percentages?
- A better day for your guests?
If a vendor cannot connect their product to one of those, you are looking at a solution in search of a problem, not a solution to a problem you have. Lize Monametsi made a related point about forecasts that never become decisions in AI Forecasting’s Decision Gap, and Jon Harmeling showed what it looks like when an operator starts from the problem in How a Chick-fil-A Operator Uses AI to Lead.
What three questions should I ask before buying restaurant AI?
1. What problem does this solve? Say it without the word AI. If the answer is vague, so will the results be.
2. What will it require of me? This is the one vendors skip. Grandiose promises still take change management, an engaged manager and an engaged owner. If guests have to download something, they have to want to. So ask what it will require of you, your managers, your staff and your guests, and force-rank those as you go through the demo. Do you have the people to implement it? Will your guests engage? Will your team adopt it?
Then ask about integration. Does it need to integrate with your POS? Do you have an open API on your POS? If you do not, none of these solutions matter. I went deeper on what separates POS platforms in Why Not All POS Systems Are Created Equal.
3. Who will own it? Like any project in your brand, if nobody owns it, it fails the moment the vendor leaves the room. Name the person, and decide who owns the outcome on day 50, day 100 and day 200, when the vendor is long gone and you may or may not still have a relationship with them.
How do I keep a new AI tool from becoming shelfware?
Treat it like any other operational project, because that is what it is. The technology is the easy part. Adoption is where it lives or dies.
- Start from the change, not the feature. Write down what will be different for your staff, your managers, your ownership and your guests if this works. If you cannot write that list, you are not ready to buy.
- Be honest about capacity. If the rollout needs a project lead, store-level training and a few weeks of a manager’s attention, make sure those exist before the contract does.
- Check the plumbing first. If the tool depends on your POS, confirm the integration and the open API before the demo, not after the signature.
- Put a name on it. One person owns the outcome, with check-ins at day 50, day 100 and day 200. When the vendor’s implementation team rolls off, that person is still there.
Why does AI on a broken process make things worse?
Because it accelerates whatever is already there. Adding AI on top of a problem that does not exist solves nothing. Adding it to an existing system when you did not really understand the problem, or had no idea how to solve it, is not a solution either. AI on top of a broken process only breaks it faster.
Before you evaluate what AI or a vendor can do, decide what needs to change for your staff, your management, your ownership and your guests. Then judge every option against that list.
Which restaurant AI vendors will win?
The ones who can be clearest about the problem they solve. A good friend of mine says it often: operators do not want AI, they want their problems solved. The vendors who understand that are the ones that will be sticky. They will get into brands, solve real problems, and become actual business partners rather than another login nobody uses.
Watch out for shiny penny syndrome. You walk a show floor, or a friend tells you about something that fixed a problem in their brand, and you want it. But their problem is not necessarily your problem. For a concrete example of AI aimed at a specific, measurable job, Eric Lam’s conversation in How AI Is Transforming Drive-Thru Performance is a good one to hear.
Frequently asked questions
Should my restaurant have an AI strategy?
Jeremy Julian argues no: AI is a tool, like a POS or an oven, not a strategy. Your strategy should be a list of business problems — no-shows, scheduling time, drive-thru speed, order accuracy, forecasting — and AI is only worth buying when a vendor can show how it solves one of them better than the alternatives.
What questions should I ask an AI vendor before buying?
Ask three. What business problem does this solve, explained without the word AI? What will it require of me, my managers, my staff and my guests, including integrations such as an open POS API? And who on my team will own it on day 50, day 100 and day 200, after the vendor has left?
What is the Tuesday test?
It is a way to judge restaurant technology by an ordinary day instead of a demo or a peak Friday night. On a slow, boring Tuesday, what does the solution change for your guests, your top-line sales, your bottom-line profit or your staff? If the answer is nothing, it is not a solution.
Why do restaurant technology projects fail?
Most fail for the same reasons, whatever the technology: nobody defined the business problem first, the change management and staff adoption were underestimated, the system could not integrate with the POS, or nobody inside the brand owned the outcome once the vendor left.
What I would do with this on Monday
- Write down the three business problems that cost you the most time or money right now. Bring that list to every vendor meeting.
- Ban the word AI from your next demo, for both sides of the table.
- Ask every vendor what their product changes on a slow Tuesday.
- Check whether your POS has an open API before you evaluate anything that needs to integrate with it.
- Name an owner for any new technology before you sign, and put day-50, day-100 and day-200 check-ins on the calendar.
If you have not already, subscribe to the monthly RTG newsletter at restauranttechnologyguys.com for every episode from the month, plus the things I have found that will help you run your restaurant better.
Thanks to our sponsors. Restaurant Technologies keeps cooking oil management off your team’s plate and out of the back lot. And Yelp for Restaurants helps operators turn discovery into covers.
