Ming-Tai Huh has spent much of his career building restaurant software.
He helped launch some of Toast's earliest online ordering and loyalty products. He later built its API and partnerships business, led its platform organization, and eventually left to run Square for Restaurants.
Then, in February, he was among the thousands of employees affected when Block cut roughly 40% of its workforce in a restructuring tied closely to AI. He wasn't the only restaurant tech executive watching AI reshape a payroll that month.
Weeks later, PAR Technology and Square both rolled out their own AI agent products for operators, joining Toast and Thanx in a field that industry press was already calling the year's defining restaurant tech trend.
Now Ming is building an AI agent for his own nine-restaurant group.
This restaurant agent is basically me, but it has infinite time and infinite capacity.
The agent can read messages, emails, and information coming from his restaurant systems. It looks for problems, identifies what needs attention, and brings the most important issues back to him.
Yet Ming doesn't think AI will make the restaurant management platform obsolete.
His view of what comes next is more practical:
- AI will remove repetitive work.
- It will connect information across more systems.
- It will help operators identify what requires attention.
- It will make existing restaurant operations software more powerful.
But someone will still need to build the interface, maintain the integrations, protect the data, and answer the phone when something breaks.
Listen to the full episode: Ming-Tai Huh on Square's 40% layoff, the restaurant tech stack, and the dream of one day quantifying the ROI of marketing
AI Was Part of Block's Restructuring, But Not the Whole Story
Block presented its layoffs as a response to how quickly AI was changing the company.
Jack Dorsey said AI tools were allowing smaller teams to accomplish more, forcing Block to reconsider how many people it needed and how the company should be organized.
Ming doesn't dismiss that explanation.
AI was already changing how teams worked
His teams were using AI to build products faster. He could see firsthand that the technology was changing how much work a smaller group of people could accomplish.
But he also saw a more familiar organizational problem.
Square and Cash App technically belonged to the same company. Operationally, they often functioned more like separate businesses, with their own teams, systems, and priorities.
They were kind of basically two separate companies, even though they were at the same board. I think Jack truly envisioned that these companies were not two companies and that they needed to be one.
That created real duplication.
AI may have accelerated the restructuring and given the announcement a clear headline. But the decision was also about bringing two large organizations together after years of rapid growth.
A layoff framed around AI is often a story about overhiring and duplicate teams wearing an AI headline.
The lesson for restaurant operators
AI will reduce some work. It will also give companies a reason to revisit processes, teams, and software they probably should have simplified years ago.
For restaurant groups, that could mean asking:
- Which reports are still being created manually?
- Which tools perform overlapping functions?
- Where are teams entering the same information twice?
- Which tasks require judgment, and which are simply repetitive?
- Which systems exist because no one has reconsidered the workflow?
AI does not automatically solve those problems. It makes them harder to ignore.
Restaurants Don't Need Another Dashboard
Ming isn't building his restaurant agent because his group lacks software.
Like most multi-unit operators, he already has plenty of it.
The POS collects sales data. Communication platforms hold messages from managers and employees. Vendors send emails. Scheduling, accounting, event, and other restaurant systems generate alerts and reports.
The problem is not access to information.
It is attention.
The operator attention problem
No owner can read every message, review every report, and investigate every unusual data point across nine restaurants.
Important information can exist inside the business without ever reaching the person who knows what to do with it.
The goal is not for the agent to run the restaurant group without Ming. It is to remove the bottleneck created by his limited time.
The agent can review far more information than Ming ever could. Because he trains it around the way he thinks, the output can come back closer to how he would evaluate the business himself.
Chowly CEO Sterling Douglass has described a similar shift in a separate conversation on this podcast: AI hasn't just made restaurant software smarter, it's made building small, custom workflows accessible to operators who never had a development team.
A single-location group Douglass worked with built its own website and deployment pipeline. A Poke Works franchisee built an automation that texts customers a free miso soup offer whenever rain is in the forecast.
Ming's agent sits in that same category: a purpose-built tool solving a problem specific to his business, not a wholesale replacement for the platforms he already runs on.
What the agent could surface
Instead of presenting another general dashboard, an effective restaurant agent could tell an operator:
- A location's labor cost has been trending upward for three weeks.
- A vendor price increase is affecting several high-volume recipes.
- The same equipment problem has appeared in multiple manager messages.
- One restaurant is consistently paying more for an ingredient than the others.
- A contract or vendor agreement is approaching a renegotiation date.
- A recurring operational complaint has not been resolved.
Restaurant operators don't need software that merely summarizes more information.
They need software that helps them understand what deserves attention and what should happen next.
From a System of Record to a System of Action
Most restaurant systems are very good at recording what already happened.
The POS stores sales. The scheduling platform stores labor. The accounting system stores expenses. The inventory platform stores purchasing and count data.
The information exists, but someone still has to interpret it.
Recording the problem is not solving it
A manager may see that food cost increased. They still have to determine whether the cause was:
- A vendor price change
- Waste or overproduction
- Inconsistent portions
- An incorrect inventory count
- A purchasing problem
- An outdated recipe cost
- Bad data entering the system
A more useful operations platform would not stop at displaying the increase.
It would help identify why it happened, show where the impact is being felt, and direct the operator toward the next decision.
That is the difference between software that holds information and software that helps a restaurant act on it. It's the same distinction meez has written about separately: a system of record stores data, while a system of action does something with it, adjusting a purchase order, flagging margin erosion, or triggering a prep alert before the problem shows up on next month's P&L. Most "AI-powered" restaurant tools today are still systems of record wearing an AI badge.
The operator should not have to search five systems just to realize there is a decision to make.
Ming's agent is intended to do the latter. It can read across different sources, recognize that something changed, and bring the issue forward while there is still time to address it.
Why AI Won't Eliminate Restaurant Software
If an AI agent can read restaurant data, identify problems, and eventually complete tasks, it is reasonable to ask what is left for traditional software to do.
Ming's answer comes down to two things:
- Accountability
- Usability
Operators are paying for accountability
"SaaS is how you take risk off the table because now someone else is accountable for the result.
Restaurant operators aren't only paying vendors for access to features.
They are paying someone else to:
- Maintain the product
- Manage the integrations
- Protect the system
- Monitor permissions
- Troubleshoot errors
- Support employees
- Fix the problem when something fails
Building an AI agent for your own restaurant group can offer enormous flexibility. But every failure becomes your problem to diagnose.
If an integration stops working, you fix it. If permissions are configured incorrectly, you own the risk. If the agent gives an employee access to information they should not see, you are responsible for the consequences.
A restaurant management platform packages that complexity into a service relationship.
AI may change what operators receive from that relationship. It doesn't eliminate the value of having someone accountable for delivering it.
Restaurant Employees Still Need an Interface
There is another practical reason restaurant software isn't disappearing.
People still need a clear way to do their jobs.
An owner may be comfortable asking an agent a question in a chat window. That doesn't mean every restaurant employee should interact with the business through the same interface.
Different jobs require different tools
A line cook needs to see a ticket and bump it from the KDS.
A chef needs to open a recipe, review the yield, and follow the production steps.
A server needs to enter an order quickly and accurately.
A manager needs to approve a schedule, review a count, or respond to an operational alert.
A line cook bumping tickets on a KDS screen isn't going to interact with a command line.
The employee doesn't care how sophisticated the AI is behind the screen. They need:
- A clear action
- A reliable workflow
- The right information
- An interface designed for the job in front of them
AI can improve what happens behind those interfaces.
It can prioritize tickets, surface prep risks, identify unusual sales patterns, or adjust recommendations using more current information.
But the interface itself still matters.
The agent can replace busywork. It doesn't replace the screen a person on the line needs to touch.
The Future of Restaurant Software Is More Powerful, Not Invisible
AI is going to change the restaurant management platform.
It will automate more analysis, review more information, and identify problems operators do not have time to find themselves.
Those are not problems restaurant operators lack the intelligence to solve.
They lack the time required to see them.
AI can help close that gap.
But restaurants will still need reliable systems. Employees will still need purpose-built interfaces. Operators will still need secure integrations. And someone will still need to be responsible when the technology fails.
AI isn’t making the restaurant management platform disappear. It is raising the standard for what that platform must deliver.
The next question is whether the restaurant’s existing systems can provide AI with the connected, trustworthy data it needs.
Continue reading: Why AI Is Forcing Restaurants to Rethink Their Tech Stack.
See how meez connects recipe, cost, and operations data across every location in your multi-unit restaurant group.
FAQ: Restaurant Management Platforms and AI
What is a restaurant management platform?
A restaurant management platform is software that supports and connects core restaurant operations. Depending on the platform, this may include point of sale, recipes, purchasing, inventory, scheduling, accounting, loyalty, reporting, and other restaurant systems.
Will AI replace restaurant management software?
AI is more likely to become embedded within restaurant management software than replace it entirely. Restaurants will still need reliable integrations, secure data, employee-facing interfaces, and vendors that are accountable when something breaks.
What are AI restaurant agents?
AI restaurant agents are tools that can review information across multiple systems, identify issues, answer operational questions, and potentially complete actions on an operator’s behalf. Their usefulness depends heavily on the quality and connectivity of the data they can access.
What is the difference between a system of record and a system of action?
A system of record stores information about what happened. A system of action uses that information to recommend or complete a next step, such as flagging margin erosion, identifying a purchasing issue, or surfacing an operational task.
Why does restaurant software still need an interface?
Restaurant employees need clear tools designed around the job they are performing. A line cook needs a kitchen display system, a chef needs access to recipes, and a server needs an ordering interface. AI can improve the intelligence behind those screens without eliminating the screens themselves.
Is SaaS still relevant when operators can build their own AI agents?
Yes. Building a custom agent gives an operator more control, but it also means they own the maintenance, security, permissions, integrations, and risk. A SaaS vendor provides both the software and accountability for keeping it operational.
