Why Mellow Mushroom Wants Its Teams Using Less Technology (And what it spent eight years building instead)

Mellow Mushroom's plan for AI is to put less technology in front of its restaurant teams, not more. It spent eight years on its data to earn that.

Ahsan Jiva is EVP, Strategy and Transformation at Mellow Mushroom. Mellow Mushroom Pizza Bakers opened in Atlanta in 1974, which makes 2026 its 52nd year. It runs about 160 restaurants across 19 states and is still privately owned. He joined in 2019 from management consulting, with no restaurant operating experience, and he now looks after strategy, digital, data, franchising and new store formats.

I built 720 Sweets from one shop to seven locations, then sold it. At Workstream's On the Clock 2026 in Atlanta, I hosted the speaker interviews. Ahsan's is the one I keep going back to, because he answered the AI question the opposite way to everyone else in the room.

Why he wants less technology in his restaurants, not more

I asked him how AI has changed the way they run the chain. Most people answer that with a list of tools.

He did not.

"I want our team members in the restaurants using technology less."

Ahsan Jiva

He knows how that sounds. His own word for it is contrarian. The common move right now is to put more technology in the restaurant and hope the team gets faster.

His objection is about attention, and the math is simple. A shift leader has a fixed number of minutes. Every minute spent reading a screen is a minute not spent with a guest.

"You're not going to interact with guests when I throw more and more tech at you," he said. Then he listed what that looks like on a real Tuesday: "Now you have a dashboard here, a dashboard there, a report here and an analysis here."

Nobody sets out to build that. It arrives one good tool at a time. Workstream's guide to the dos and donts of restaurant technology makes the same point from the other side. Every new system takes a piece of someone's attention, and almost nobody adds that cost up.

The new manager app gives an action, not a chart

Mellow Mushroom is building tools on top of its own data. One of them is a new app for general managers.

Listen to how he describes it. He talks about what it removes, not what it adds.

"We want to get them out of the office, out of their desktops, out of dashboards," he said. Out of trying to work out what the data is telling them.

The app does not hand a manager a chart. It hands them a recommended action. And it expects the manager to argue with it.

"We want to give them the recommended action, but then have them use their own experience," he said. Maybe the action needs to be a little different. That is the design, not a flaw in it.

"They can spend less time with the technology and really get back to talking to guests."

Ahsan Jiva

That is the whole test. Not how clever the model is. Whether the manager is on the floor more this month than last month.

Eight years on the data, and a team of two

This is the part that makes the position more than a slogan. He did not simply stop buying tools. He spent years making the data good enough to answer questions on its own.

Mellow Mushroom has been investing in a modern data stack for seven to eight years. He is honest about why. Maybe they saw the AI wave coming. Maybe it was luck and timing. Either way the data is in a good place now, and that is what lets them build their own tools.

"We're doing it internally with a team of two people."

Ahsan Jiva

Two. Not a department. He contrasts it with how this used to work, when anything ambitious meant you "needed SWAT teams of engineers."

What changed is that his small team can now write software by describing what they want, with a lot of human review, because everyone knows how that can go wrong. The coding stopped being the hard part. The data work is the hard part, and that took eight years.

You should not need eight years to get your records straight. A manager digging through folders for a signed form or a start date is a manager who is not with the team. Workstream's HR platform keeps every employee record, document and onboarding step in one place from the first hire, so the answer is already there. Explore Workstream →

Buy from open partners, and name your data properly

I asked what an operator who has not started yet should be recording now. His answer had nothing to do with AI.

Start with who you buy from.

"You have to work with partners that are open."

Ahsan Jiva

Open means open systems and open APIs. It also means scalable and adaptable, so the system can grow and change as the business does. And it means they want to listen, which he rates highest of the four. He calls them vendor partners, because he thinks a supplier who will not act on what the restaurants tell them is not a partner.

"If there doesn't seem to be a sense of urgency around that, then you're not a good partner to me," he said. He has ended relationships over it.

The second half of his answer is the least glamorous thing anyone said to me all week, and it is the one I would act on first. How you name a field. How you name a data point. Get that wrong and everything you build on top of it is wasted.

You can have all the great data in the world, he said, but the models cannot read it. His phrase for the fix is being hyper specific about the little things. He says the same about data maturity: know how far along you really are before you promise anyone an answer.

Hire for curiosity, not for an AI resume

He hired his data role in 2022, before most people had heard of ChatGPT and Claude. So he could not have hired for AI skills. He had to hire for something else.

"I don't think it's necessarily about hiring an AI engineer," he said. Some firms need one for a specific reason. Most do not.

"You just need people that are smart, dedicated and have a good work ethic and are curious."

Ahsan Jiva

Then four words that sum up the whole hiring policy: "You'll figure it out." His view is that nobody was born an expert in this, the tools themselves teach you, and a curious person keeps exploring until the answer shows up. Workstream's list of restaurant interview questions is a decent place to start if you want to test for that instead of scanning resumes for the word AI.

Inside the restaurants he treats the same idea as part of the culture, not a perk. Learning and development sits at the core, because a company that keeps teaching its people new skills ends up with better people. Sometimes the skill is data. Sometimes it is how to talk to a stranger.

"Some of our younger members in our restaurants struggle with that, because they've grown up with phones in their faces," he said. That is the same argument he makes about the floor, turned on his own team. Time on a screen is time not spent talking to people.

What this means if you run 2 locations, not 160

You do not have a two person data team or eight years to spend. The approach still works at your size, because it is not about the software.

• Count the screens before you add one. Ask your manager how many logins a normal Monday takes. If that number is already high, the next tool has to replace one.
• Buy from people who answer you. Open systems matter. A supplier who acts on your feedback matters more.
• Fix your names now. Spell your items, roles and locations the same way in every system. It costs nothing today and saves a cleanup job later.
• Hire the curious one. Tools change. The habit of working things out does not.
• Judge every tool by time on the floor. If a tool does not put a manager in front of guests more often, it failed, however good the dashboard is.

I asked him whether this is a people business. His answer was quick. "If anyone in the restaurant industry tells you otherwise, that's a red flag to me." Then came the line I would put on the office wall. "If you take care of the people, they will take care of the guests."

He started on his own back office. Mellow Mushroom runs Workstream in its company owned restaurants, and franchisees who are already on another platform keep their flexibility. He is blunt about what hiring and onboarding used to feel like. "It was painful. It was very analog. It was very disjointed systems, a lot of double entry." If that sounds like your week, Workstream's piece on reducing admin burnout describes the same problem.

Frequently Asked Questions

Should a restaurant add more technology to the floor or less?

Less, if the goal is guest experience, and that is the position Mellow Mushroom takes across 160 restaurants. Ahsan Jiva wants team members using technology less so they spend that time with guests. His test for a tool is whether a manager ends up on the floor more often, not whether it has more features.

How many people does it take to build a restaurant data stack?

Two, in Mellow Mushroom's case, after about eight years of data work. Ahsan Jiva says the company has been investing in a modern data stack for seven to eight years, and now builds its own tools with a team of two. The small team works because the data came first, not because software got easy.

What should an operator look for in a technology partner?

Look for four things: open systems with open APIs, scale, adaptability, and a partner who acts on your feedback. Ahsan Jiva rates the last one highest and has ended supplier relationships when there was no sense of urgency about an operating problem. He calls them vendor partners because a supplier who cannot change is not a partner.

Why does naming your data matter for AI?

Badly named fields make good data unreadable to the software you want to use. Ahsan Jiva says you can have all the great data in the world and the models still cannot read it. His advice is to be hyper specific about how each field is named. He also says to be honest about how mature your data really is.

Do you need to hire an AI specialist to use AI in a restaurant?

Most restaurant companies do not, and Mellow Mushroom hired its data role in 2022, before the current tools existed. Ahsan Jiva hires for critical thinking, work ethic and curiosity rather than for a specialist resume. His reasoning is that the tools teach the user, so a curious person will work it out while a credential goes out of date.

Request a demo of Workstream →

About the author. Wilson K Lee built 720 Sweets from one shop to seven locations and sold it. He now runs wilsonklee.com, where he publishes real numbers and playbooks for food and beverage owners, and he hosted the speaker interviews at Workstream's On the Clock 2026.