In July 2024, Chris Caldwell took over as chief information officer of Brinker International, the parent company of Chili’s Grill & Bar. Over the next two and a half years, he made a decision that bucks the current tech obsession: instead of chasing the latest artificial intelligence models, Caldwell invested in 23,000 iPads for servers and a complete overhaul of Wi-Fi infrastructure across 1,200 U.S. restaurants. The result has been a steady improvement in same-store sales, driven by faster service, fewer order errors, and a more reliable technology backbone.

The Hardware and the Overhaul

When Caldwell arrived at Brinker, he encountered a restaurant chain experimenting with flashy tech—robots that bussed tables and sang “Happy Birthday”—while the foundational systems creaked. “The first thing I heard was that we’re doing a lot of state-of-the-art things, but the foundational systems were really having some challenges,” he told Business Insider. Servers didn’t always have enough tablets to take orders, and when they did, the devices sometimes failed to transmit orders to the kitchen reliably.

The fix was deliberate and large-scale. Brinker deployed approximately 23,000 iPads as the primary order-taking tool for waitstaff. The software on these tablets was redesigned to mimic the simplicity of consumer apps, reducing training time for new employees and minimizing order-entry mistakes during peak hours. Meanwhile, Wi-Fi access points were replaced in every Chili’s location, and bandwidth was boosted to ensure that operational traffic—orders, payments, kitchen displays—never competed with guest Wi-Fi.

“I’m not chasing AI to go, ‘Guys, Anthropic just released a new LLM, and we need to use it because it’s cool,’” Caldwell said. Instead, the focus was on making life easier for restaurant teams and customers.

Why This Matters for Anyone Running Tech

There is a lesson here that extends far beyond the restaurant industry. For IT leaders, business owners, and even Windows admins overseeing device fleets, the Chili’s approach is a reminder that AI is only as good as the platform it runs on.

For restaurant operators and service businesses: Investing in reliable handheld devices and robust wireless networks directly improves table-turn times, order accuracy, and employee morale. A server who can take an order tableside without walking to a fixed terminal saves minutes per table. Over hundreds of covers a night, that adds up to real revenue. If your network drops, the entire operation freezes—no payment processing, no kitchen tickets, no happy guests.

For enterprise IT decision-makers: Before greenlighting an AI pilot, ask whether your network can handle the additional data load, whether your endpoints are manageable, and whether your frontline applications are intuitive enough that employees actually use them. A generative AI assistant for inventory forecasting is useless if the point-of-sale system crashes at dinner rush.

For Windows administrators and device managers: Chili’s chose iPads, but the management challenges are identical to what you face with Windows-based kiosks, ruggedized handhelds, or shared PCs. You still need enrollment, configuration, updates, security policies, and remote wipe capabilities. Brinker had to solve for shift changes, device breakage, and lost tablets—all of which require the same discipline as managing a fleet of laptops. The operating system may differ, but the enterprise headaches are universal.

For developers of frontline software: The order-taking app redesign at Chili’s prioritized speed and simplicity. That is not a trivial goal. Employees compare work tools to the apps on their own phones; if your interface is clunky, they will find workarounds, or worse, make mistakes. Reducing the number of taps to modify an order or process a payment can be the difference between a five-star review and a complaint.

How We Got Here: A Timeline of Restaurant Tech Hype and Reality

The restaurant industry has been awash in AI buzz. Drive-thru voice agents at Starbucks and Burger King, kitchen forecasting algorithms, chatbot ordering systems, and computer vision for quality control have all grabbed headlines. The promise is tantalizing: automate repetitive tasks, predict demand, and free up humans for higher-value work.

But the reality is messier. Voice agents still stumble over accents and background noise. Chatbots can infuriate a loyal customer who just wants to speak to a manager. Predictive models feed on clean data, but if your order-entry system is inconsistent, those forecasts are garbage.

Chili’s itself had dabbled in novelty—robot assistants that cleared plates and performed for guests. Those experiments generated social media buzz but did little for throughput or employee satisfaction. When Caldwell started talking to restaurant teams, the message was clear: fix the basics. Orders needed to reach the kitchen faster. Tablets needed to be available. The Wi-Fi had to stay up.

This pattern is not unique to restaurants. In retail, hospitals, and logistics, the organizations that wring the most value from technology are often those that first modernize the unglamorous plumbing: network infrastructure, device management, identity systems, and application performance. Without that, AI becomes another layer of complexity on a shaky foundation.

What You Should Do Now

If Chili’s strategy resonates, here are concrete steps for your own organization—whether you run a chain of coffee shops or manage IT for a manufacturing floor.

1. Audit your operational infrastructure
Walk through a frontline worker’s day. How many times do they wait for a system to load? How often does an order get lost? Is the Wi-Fi signal strong in every corner where work happens? Document every friction point.

2. Prioritize reliability over novelty
Before adding an AI feature, ensure your existing tech stack is stable. That means redundant internet connections, consistent device configurations, and proactive monitoring. A single network outage can undo months of efficiency gains.

3. Invest in user experience for frontline apps
The people taking orders, stocking shelves, or checking in patients are not IT professionals. Design interfaces that require minimal training and are forgiving of errors. Use real-time validation and simple workflows. If possible, test designs with actual employees during a busy shift.

4. Treat mobile devices as managed endpoints
Whether you deploy iPads, Android tablets, or Windows handhelds, they need the same rigor as office PCs. Use a mobile device management (MDM) solution to enforce security policies, push updates, and track inventory. Plan for breakage and theft. Have spares ready.

5. Segment your network
Guest Wi-Fi should never compete with business-critical traffic. Create separate VLANs for point-of-sale systems, handheld devices, and customer access. Prioritize operational data with quality-of-service (QoS) rules.

6. Measure what matters
Don’t judge success by the number of tablets deployed. Track order-entry time, kitchen ticket accuracy, employee training duration, and guest satisfaction scores. These operational metrics reveal whether the technology is actually making a difference.

7. Use AI selectively, with a clear problem in mind
Caldwell and his team are exploring AI for specific use cases, such as estimating when an online order will be ready for pickup. That is a concrete, measurable problem. Start with something similar in your own domain: a repetitive forecasting task, a documentation shortcut, or a data-aggregation chore. Then measure whether it saves time or reduces errors.

The Outlook: A Smarter Restaurant, One Stable Network at a Time

Chili’s turnaround is ongoing, and the chain has already reported sustained comparable-sales growth—a metric that management attributes to better food, service, atmosphere, and value. Technology is not the sole driver, but it is an essential enabler. When servers aren’t fighting with dead tablets or spotty Wi-Fi, they can focus on guests. When orders flow seamlessly to the kitchen, food arrives faster and more accurately. When payments are processed tableside, customers leave happier.

Caldwell does use AI, by the way. Brinker corporate employees have access to Microsoft Copilot, and the CIO himself brainstorms with chatbots to pressure-test strategic decisions. But that corporate use is separate from restaurant operations, and it is deployed where the risks are lower and the upside is clearer.

Over the next few years, Chili’s will likely layer more intelligence onto its now-stable foundation. Better pickup-time estimates are just one possibility. Predictive staffing, smarter menu recommendations, and even computer vision for kitchen quality control could follow. But those projects will succeed because the packets are flowing and the devices are reliable.

For everyone else, the message is clear: before you ask what AI can do for your business, ask whether your business can support AI. If the answer is no, it’s time to invest in the unglamorous but essential work of making technology dependable. In the end, a working tablet and a strong Wi-Fi signal may be more transformative than any model.