If Maddy Shannon could give every restaurant operator one superpower, it wouldn't be the ability to collect more data. It would be the ability to know which question actually matters.

While building a product called Profit AI at SpotOn, Shannon began asking operators a deceptively simple question. If they had a genie in a bottle, one that could answer anything about their business, what would they ask?

Again and again, the room would go quiet.

That pause fascinated her. Not because operators lacked information, but because they were surrounded by it. Their POS systems, invoices, labor reports, and inventory tools generated more numbers than they could realistically process. The challenge wasn't finding data. It was knowing where to focus their attention.

That distinction sits at the center of today's conversation around restaurant AI insights. The most valuable tools won't be the ones that surface more reports or build more dashboards. They'll be the ones that help operators understand what deserves their attention, and what action to take next.

"Data is actually plentiful. There's a lot of it. The issue that restaurants have always had, and tech serving restaurants has always had, is attention. Attention is very scarce."

Shannon arrived at this perspective from both sides of the industry. Before moving into product strategy, she worked on the business teams at Alinea and SingleThread, which gave her firsthand experience with the realities of restaurant operations.

On a recent episode of The meez Podcast with founder Josh Sharkey, she argued that AI's biggest opportunity has little to do with chat interfaces or flashy technology. Instead, it's about protecting the scarcest resource every operator has: time and attention.

Her framework reframes the conversation entirely. Rather than asking "which AI tool should we buy?", operators should ask:

• What decision am I actually trying to make?
• What information would let me make that decision in five minutes?
• Is this tool giving me an answer, or simply another place to look?

Those questions reveal the difference between technology that creates work and technology that removes it.

Listen to the full episode: Maddy Shannon on 'the Anticipate and deliver' approach, the price of a restaurant meal, and the Ai paradigm

Restaurants Are Drowning in Data and Starved for Attention

Shannon's observation that data is plentiful and attention is scarce captures a reality operators live with every day.

Very few restaurants struggle to generate information. Sales reports, labor costs, purchasing data, inventory, guest feedback, and reservations are constantly flowing into the business. The challenge is turning all of those disconnected signals into something useful before service begins.

Industry research reflects exactly that. FSR magazine describes restaurant operators as swimming in data, yet much of it stays trapped inside legacy systems that don't communicate with one another. The result is a slow, frustrating process of piecing information together before anyone can make a meaningful decision. In today's competitive environment, the difference between surviving and thriving largely depends on the ability to transform data into decisions that improve guest experience and operations.

That's the entire ballgame. Collecting more numbers has never been the constraint.

Modern Restaurant Management reached a similar conclusion in its 2026 outlook. The industry's next step isn't gathering additional information; it's moving from simply collecting data to actually using it, because many restaurants struggle to apply what they gather in ways that meaningfully change the business.

That's an important distinction. An operator juggling fifteen dashboards isn't necessarily informed. More often, they're overwhelmed. Every report competes for limited attention, making it harder, not easier, to identify what deserves immediate action.

For Shannon, that's the real bottleneck AI should solve.

The Answer Isn't a Better Dashboard. It's a Diagnosis.

Shannon doesn't believe AI's job is to build prettier reports. She believes its job is to interpret them.

During the podcast, she compared the experience to visiting a doctor. When you aren't feeling well, you don't expect to perform the diagnosis yourself by reviewing lab results one organ at a time. You expect the doctor to evaluate everything they already know and tell you what's actually wrong. Restaurant software, she argues, should work the same way.

"It's diagnose me. You have all the information. You've looked at me. You tell me what's wrong. And that's, I think, where we're going."

Josh Sharkey hears the same request from operators. Especially among sophisticated multi-unit groups, he says the conversation has shifted beyond interface design.

Attractive dashboards and polished visualizations are appreciated, but they're no longer enough. Operators don't want another screen to interpret. They want software that tells them what changed, why it matters, and where they should focus this week.

That's the difference between reporting and recommendation.

A dashboard that shows food cost rising from 22% to 24% tells you what happened. A system that identifies the recipe responsible, explains why margins changed, and estimates the annual savings from a specific ingredient substitution gives you something far more valuable: a decision.

That's the philosophy behind meez's menu engineering matrix. Rather than forcing operators to dig through disconnected reports, it organizes menu items according to profitability and popularity so opportunities become visible immediately.

The objective isn't simply to surface more information. It's to reduce the work required to act on it, which is what separates menu intelligence that reports from menu intelligence that recommends.

Answers Are Only as Good as the Recipe Data Underneath Them

Of course, there's an important caveat. AI can only be as trustworthy as the data it's built on.

Sharkey raised this point directly during the conversation, arguing that clean, structured data is one of the most overlooked ingredients in effective AI. Without it, even sophisticated models produce recommendations that sound convincing while pointing operators in the wrong direction.

An AI answer built on stale recipe data is still the wrong answer, just delivered with greater confidence.

That's one reason restaurant AI often disappoints.

The technology itself may be capable. But if ingredient costs haven't been updated after supplier price increases, recipes aren't standardized, or menu items aren't mapped consistently across locations, the system is making recommendations from an incomplete picture.

As meez explains in its breakdown of AI-washed versus AI-real tools, recipe data sits at the center of nearly every operational insight an AI system generates. Food costs, yields, prep loss, menu pricing, and profitability all trace back to the quality of that underlying information.

The broader industry is arriving at the same conclusion. 

Modern Restaurant Management's 2026 technology forecast argues that restaurants seeing measurable returns from AI are also the ones investing in disciplined data management, because operators pull information from POS, invoices, payroll and reservations, but most of those systems speak different languages, and AI needs clean, consistent data to be useful. 

For kitchens, that foundation starts with recipes. When recipes include accurate yields, unit conversions, prep loss, and current ingredient costs, every downstream insight becomes more reliable. When they don't, no amount of artificial intelligence can compensate. It's why engineering a menu directly from structured recipe data produces stronger recommendations than layering analytics on top of disconnected spreadsheets after the fact.

The intelligence isn't created by AI alone. It's created by giving AI something accurate to reason from.

This Is a Multi-Unit Problem Before It's an AI Problem

The attention problem doesn't stay the same as a restaurant grows. It compounds.

A chef running a single location can often rely on instinct and familiarity. They know which menu items have been slipping, which vendor prices changed last week, and where service consistently breaks down. Much of that knowledge lives in their head.

Scale changes everything. A director of operations responsible for a dozen locations can't personally reconcile every recipe change, food cost fluctuation, labor report, and sales trend across the business. The volume of information becomes impossible to manage manually, and every additional location widens the gap between collecting data and acting on it.

That's why Shannon's point about attention is really a point about growth. As restaurants expand, the challenge isn't generating more information. It's helping operators identify what deserves their attention without forcing them to dig through another stack of reports. That's where restaurant operational intelligence becomes valuable, not because it produces more analytics, but because it consolidates scattered signals into a picture operators can actually use.

The financial reality makes those decisions even more consequential. Restaurant Business reported that full-service restaurant margins hovered around 2.8% last year, while both food and labor costs have risen roughly 35% since 2019. At margins that thin, delayed decisions become expensive decisions. And the spread between operators is widening.

One industry analyst told Modern Restaurant Management that the real story of 2026 is that the gap between great operators and average operators has never been wider, and it's the year AI needs to move from buzzword to demonstrated line-item savings.

The operators pulling ahead aren't the ones with the largest collection of dashboards. They're the ones whose systems shorten the distance between data and action.

The Market Is Already Moving Toward Answers

You can see Shannon's thinking reflected well beyond one podcast conversation.

Walk the floor of a restaurant technology trade show today, and the conversation has noticeably shifted. The latest wave of AI products isn't promising another analytics platform or reporting dashboard. Instead, vendors are introducing assistants designed to sit on top of existing restaurant data and surface recommendations automatically.

Reporting from the 2026 National Restaurant Association Show highlighted exactly this trend, describing assistants that sit on top of a restaurant's data and provide insights and action items, rather than additional reports.

That's the whole shift in one sentence: from where's the data to here's what to do.

The move matters because adoption is still relatively early. According to the National Restaurant Association's 2026 report, only about 26% of restaurant operators say they're currently using AI-related tools. That leaves a significant majority still figuring out how AI fits into their operations, and it creates a real opening for operators who build the right foundation now.

But Shannon and Sharkey both caution against confusing early adoption with effective adoption. Restaurants that rush toward AI without first organizing the underlying data are likely to generate recommendations they can't trust. A sophisticated model doesn't solve inconsistent recipes, disconnected systems, or outdated costs.

That's why menu engineering has become more than an occasional pricing exercise. Done well, it's an ongoing discipline that continuously improves the quality of the data feeding every recommendation. AI simply amplifies the quality of whatever foundation already exists.

The Operator Still Decides. AI Just Removes the Digging.

For all the excitement around AI, Shannon is careful not to overstate its role. She isn't arguing that software should replace operators. She's arguing that it should eliminate unnecessary work.

Even the most sophisticated AI can only interpret the information it has access to. It can't account for a conversation with a landlord, a staffing challenge, changing neighborhood dynamics, or the intuition that comes from years of running a restaurant. Those decisions remain fundamentally human. What AI can do is dramatically shorten the time it takes to reach them.

Instead of spending hours piecing together reports from different systems, operators can begin with a concise explanation of what changed, why it matters, and where the biggest opportunity lies. The technology surfaces the signal. The operator decides whether, and how, to act on it. Sharkey sees that division of responsibility as essential. The software compresses hours of investigation into a five-minute read. The judgment stays with the operator.

That perspective brings the conversation back to Shannon's original genie question. The silence she encountered wasn't evidence that operators lacked answers. It was evidence that they were overwhelmed by possibilities. When every report presents another metric to investigate, even knowing where to begin becomes difficult.

The right AI doesn't eliminate complexity by making decisions for operators. It removes the burden of searching through the complexity. Once recipe data is clean, structured, and connected, good AI menu optimization can surface the questions that matter, and increasingly, the answers alongside them. The pause gets shorter. Eventually, it disappears.

The best restaurant AI insight isn't the one that shows you the most. It's the one that tells you what matters next.

See how meez turns clean recipe data into decisions your team can act on at every location. Explore meez for operations leaders.

FAQ

What are restaurant AI insights?

Restaurant AI insights are recommendations and action items an AI system surfaces by analyzing a restaurant's operational data, such as recipe costs, sales performance, and invoices. The most valuable tools don't simply display more metrics; they interpret the information and surface what deserves attention. As SpotOn's Maddy Shannon frames it, the goal is to deliver answers that respect an operator's scarcest resource: attention.

Why do restaurants collect so much data but struggle to use it?

The problem isn't collecting data, it's connecting and acting on it. Most restaurants already have enormous amounts of operational information spread across POS systems, payroll platforms, inventory software, invoices, and reservations that don't communicate with one another, so turning those disconnected points into a clear decision can take hours operators don't have. As FSR magazine notes, the restaurants separating themselves today aren't gathering more data; they're getting better at turning existing data into timely decisions.

How does AI actually help with menu profitability?

Effective AI menu optimization connects recipe costs, sales performance, and profitability to explain not just what changed, but why. Instead of telling an operator that food cost increased, a well-designed system identifies the recipe driving the change, highlights the ingredient responsible, and estimates the financial impact of a specific adjustment. That turns operational reporting into actionable menu decisions rather than another metric to interpret.

Can AI give restaurants wrong answers?

Yes, and it's one of the biggest risks. AI can only work with the information it's given. If recipe costs are outdated, ingredient prices haven't been updated, or menu items aren't mapped consistently across locations, the recommendations may sound authoritative while pointing operators in the wrong direction. As meez explains in its guide to AI-washed versus AI-real tools, trustworthy AI begins with trustworthy recipe data.

Why is clean recipe data important for restaurant AI?

Recipe data underpins nearly every operational insight AI produces. Food costs, yields, prep loss, portion sizes, and pricing all originate with recipes, so when that information is accurate and consistently maintained, every downstream recommendation becomes more reliable. Modern Restaurant Management reports that restaurants seeing meaningful AI results are also the ones investing in clean, consistent data across their systems.

Do multi-unit restaurants need AI insights more than single locations?

Multi-unit operators tend to benefit the most because the attention problem grows with every new location. A single restaurant can often rely on institutional knowledge and hands-on oversight, but an operator responsible for multiple locations can't personally reconcile every sales report, recipe update, labor trend, and purchasing change. Restaurant operational intelligence becomes valuable because it consolidates those signals into a short list of priorities. With full-service margins around 2.8%, even small delays in identifying problems can add up across a portfolio.

How many restaurants are actually using AI right now?

About 26% of operators reported using AI-related tools, according to the National Restaurant Association's State of the Restaurant Industry 2026 report. Adoption is still relatively early, which creates an opportunity for operators who establish strong data foundations now. But the advantage comes from pairing AI with reliable operational data, not from adopting AI for its own sake.

Should AI make decisions for restaurant operators?

No. Both Maddy Shannon and Josh Sharkey emphasize that AI should support judgment, not replace it. Operators bring context software can't fully understand, from a changing neighborhood to staffing realities, vendor relationships, and intuition built from years of experience. AI's role is to eliminate the hours spent digging through reports, highlight what matters most, and suggest possible actions. The final decision still belongs to the operator.