Soon the customer placing your next order may not be a person at all. It may be their AI agent. The restaurants that win will not have the flashiest AI. They will have the cleanest data.
Key takeaways
- Agentic commerce is when an AI agent finds, customizes, and places an order on a guest's behalf, instead of the guest tapping through your app or menu.
- It is not a far-off idea. Google has launched a food ordering protocol, Papa Johns is an early adopter, and analysts project trillions in agentic commerce value within a few years.
- The common takeaway from early stumbles, like the ChatGPT checkout pullback, is that agentic commerce is hype. That is the wrong lesson. The barrier is infrastructure, not demand.
- Here is the unique point: your menu is becoming an API. The brands that win will have the cleanest data and the most open, connected systems, not the flashiest chatbot.
- NCR Voyix runs an API-first, edge-native platform that supports roughly $1.4 trillion in transaction volume, which is the kind of foundation agentic ordering will require.
In late 2025, one of the biggest names in AI let people buy products straight from a chat window. A few months later, it quietly pulled the feature back. The headlines wrote themselves. Agentic commerce, many concluded, was overhyped and not ready.
That is the wrong lesson. The pullback did not happen because people do not want AI to help them buy things. It happened because connecting real transactions to real merchant systems is hard. Menus go stale. Data sits in silos. Systems cannot talk to each other cleanly. In other words, the barrier was never demand. It was infrastructure.
For restaurants, that distinction matters, because agentic commerce is already arriving at your door, whether or not your systems are ready for it.
The wrong lesson from the ChatGPT checkout stumble
- Agentic commerce is not a feature you bolt on. It is an infrastructure problem you solve.
- Agentic commerce is not failing because guests will not use it. It is difficult because real ordering depends on clean menu data, live availability, payment, routing, and fulfillment. The challenge is not the AI interface. It is the infrastructure behind it.
What is agentic commerce for restaurants?
Agentic commerce means a software agent acts like a personal assistant for the guest. Instead of a person opening an app, searching the menu, reading every modifier, and checking out step by step, the agent does that work for them.
Picture a guest telling an assistant: “Order dinner for four from somewhere nearby, avoid peanuts, and reorder the sides we liked last time.” The agent finds the restaurant, checks availability, builds the order, applies the right offer, and pays. The guest never opens your app.
This is different from the AI most restaurants already use. A drive-thru voice bot or a website chatbot is reactive. It waits for input and responds. An agent is proactive. It observes, decides, and acts on the guest's behalf, within limits the guest sets. That shift, from reactive to autonomous, is the real change.
How the order flow changes
What the data says
Agentic commerce by the numbers
- Trillions in value. Global agentic commerce transaction value is projected to reach several trillion dollars within a few years, and by 2030 up to one in ten US retail dollars could flow through AI agents. (Juniper Research; McKinsey)
- Fast growth off a small base. In restaurants, agentic ordering grew from about 1% of orders in 2023 to an estimated 3% in 2024, and is accelerating as protocols mature. (Industry estimates)
- Trust is still being earned. Only about 18% of US consumers currently express interest in AI menu recommendations, and nearly 30% feel technology can detract from dining, so autonomous ordering will follow demonstrated trust, not precede it. (Mintel)
Why is agentic commerce happening now?
Because the underlying infrastructure is being built in public, and it is moving quickly. A few signals from just the last year:
- Open standards have arrived. Google introduced a Universal Commerce Protocol that now extends to food ordering, and other major AI players have published their own agentic checkout specifications. These are the rails that agent ordering will run on.
- Major brands are already live. Papa Johns became an early partner for Google Cloud’s food ordering agent, deploying unified voice and text ordering across app, web, phone, and kiosk. This is no longer a lab demo.
- The market math is large. Independent analysts project agentic commerce transaction value in the trillions of dollars globally within a few years, with a meaningful share of restaurant and retail orders flowing through agents by 2030.
- Costs favor it. Early industry estimates put agentic orders at roughly one dollar per order, and the model lowers cost for both guests and brands, creating the kind of tailwind that can accelerate adoption.
If agentic ordering depends on real-time data, high-volume transactions, and reliable integrations, then scale matters. This gives NCR Voyix a meaningful vantage point.
- NCR Voyix processes roughly $1.4 trillion in annual transaction volume
- Approximately 110 million mobile orders and 72 million loyalty transactions a month run through the Voyix Commerce Platform, a sign of how much ordering already flows through connected, machine-readable channels.
The shift most restaurants are missing: your menu is becoming an API
For twenty years, the digital front door of a restaurant was a screen a human looked at. You optimized the app, the website, the kiosk, the drive-thru display. The whole game was making that screen easy for a person to use.
Agentic commerce changes the audience. Increasingly, the thing reading your menu is not a person. It is an agent. Agents do not care how pretty your app is. They care whether your menu, prices, modifiers, and availability are accurate, structured, and available through an open connection. In practical terms, your menu is becoming an API.
That is why the winners of this era will not be decided by who has the flashiest chatbot. They will be decided by who has the cleanest data and the most open, connected systems. If an item is out of stock in the kitchen, the agent needs to know instantly. If a price changes, the agent needs the new one. If a guest has a preference, the system needs to hold it. This is an operations and architecture question far more than a marketing one.
From my vantage point
From my vantage point working with restaurant brands, the AI conversation often starts in the wrong place. Operators ask which chatbot, voice assistant, or agent they should be watching. The better question is whether their data, ordering, loyalty, and kitchen systems are ready for any agent that comes next.
Agentic commerce will not be won at the interface. It will be won in the operating model behind it. If a menu item is unavailable, the agent needs to know. If a price changes, the agent needs the current price. If a guest has a preference, the system needs to recognize it across channels. That level of readiness does not come from one new AI tool. It comes from clean data, open connections, and a platform built to support real-time operations.
That is where I believe restaurant leaders should focus now. Do not try to predict which agent will win. Build the operational readiness so your brand is prepared to show up accurately, reliably, and consistently wherever your guests choose to order.
Challenger point: do not chase the chatbot, fix the foundation
- It is tempting to answer agentic commerce by buying a shiny consumer-facing AI. That is chasing the wrong thing. A voice bot on a broken data layer just automates the chaos faster.
- The durable advantage is underneath: a single source of truth for menu, pricing, and availability, open APIs that approved agents can use, and systems resilient enough to keep syncing during a rush.
The good news is that restaurants do not need to wait for the agentic commerce market to settle before taking action. The same work that prepares a brand for AI ordering also improves today's digital ordering, loyalty, operations, and guest experience.
How can restaurants get ready for agentic commerce?
You do not need to predict which agent or platform will win. You need an agent-ready commerce environment that can support any of them. Four moves matter most:
- Clean up your data. Make menu, pricing, modifiers, and availability accurate and consistent across every channel. This is the single highest-value step.
- Make your systems open. Favor an API-first platform so approved agents and partners can read your menu and place orders without custom, brittle integrations.
- Unify your guest data. Agents personalize based on preferences and history. A connected view of the guest, across loyalty and transactions, is what makes agentic ordering feel relevant instead of generic.
- Keep the kitchen in sync. An agent order is only as good as your ability to fulfill it. Real-time availability and clean order routing keep agent demand from breaking the line.
What an agent-ready platform needs to support
- An agent-ready platform needs clean data, open connections, reliable order routing, and real-time operational awareness. For NCR Voyix, that means building interoperable capabilities into the platform rather than treating agentic commerce as a bolt-on.
- The Voyix Commerce Platform is API-first and edge-native, with a unified data foundation across POS, loyalty, and operations, and support for open commerce protocols as they mature. Our view is simple: help operators prepare for autonomous, agent-driven ordering while keeping people firmly in control of the guardrails.
The takeaway for restaurant leaders
Agentic commerce will not arrive as a single dramatic moment. It will accumulate, one behavior and one protocol at a time, until ordering through an agent feels normal. The brands that thrive will be the ones that used this moment to make their data more trustworthy, their systems more connected, and their operations more resilient.
You cannot control which AI agent your guest chooses. You can control whether your restaurant is ready when that agent comes knocking. In the agentic era, visibility will favor the brands whose data can be trusted, whose systems can connect, and whose operations can deliver.
Get the foundation ready for what is next.
Explore the NCR Voyix restaurant platform, see how unified guest data powers smarter engagement, or get in touch with our team to talk through an agent-ready roadmap for your brand.
Frequently asked questions
What is agentic commerce for restaurants?
- Agentic commerce is when an AI agent finds, customizes, and places a food order on a guest's behalf, rather than the guest tapping through an app or menu themselves. The agent handles discovery, modifiers, offers, and payment, interacting directly with the restaurant's ordering systems.
How is agentic ordering different from a chatbot or voice bot?
- A chatbot or voice bot is reactive. It waits for input and responds. An agent is proactive and autonomous. It observes context, makes decisions, and completes an order on the guest's behalf within limits the guest sets, moving from answering questions to taking action.
Is agentic commerce actually happening in restaurants yet?
- Yes, in early form. Google has launched a food ordering protocol, major brands like Papa Johns are live with AI ordering agents, and analysts project trillions of dollars in agentic commerce value within a few years. Adoption is early but accelerating as open standards mature.
What do restaurants need to do to prepare for agentic commerce?
- Focus on the foundation: clean and consistent menu, pricing, and availability data across channels; open, API-first systems that approved agents can connect to; unified guest data for personalization; and real-time kitchen sync so agent orders can be fulfilled reliably.
Will AI agents replace restaurant apps and staff?
- No. Agentic commerce adds a new way for orders to reach a restaurant, alongside apps, web, and in-store ordering. Human trust is still being earned, and people remain in control of the guardrails. The near-term impact is on the ordering channel and the data behind it, not on replacing staff.
About the author
Ed Jantz — Vice President, Restaurant Product & Design, NCR Voyix
Ed leads restaurant product and design at NCR Voyix, where he focuses on how modern platform architecture helps operators run and scale their businesses. He works closely with multi-unit and enterprise brands on the technology decisions that shape speed of service, resilience, and growth, and is a frequent voice on the future of restaurant technology.
Share this post
Related blog articles
Even more expert insights into important retail and restaurant trends.
How retailers should split workloads between cloud and edge for real-time store operations
How to reduce shrink at self-checkout with visual AI
Restaurant & Retail Tech: Why Cloud‑to‑Edge Architecture Is Becoming the New Standard
