At RestaurantSpaces Spring in Miami, a group of franchise development and construction leaders gathered for a behind-closed-doors conversation on AI — not the keynote version, but the real one. What tools are people actually using? What's working, what's not, and how do you even know where to start?
The room got honest fast.
Before anyone got into specific tools, one participant reframed the whole conversation: the biggest mistake people make with generative AI is assuming it's being straight with them.
"We treat it like a hostile witness," said a rep from one brand. "We assume it's lying to us. It's built to please you the same way social media is — you ask it something, it gives you an answer, you probe a little deeper, and it immediately says, 'Oh no, you're right, I was wrong.' So we use it for objective, factual tasks. Data spreadsheets. Administrative work. Not creative thinking — we've found it too unreliable for that."
Others in the room have landed in a similar place. The emerging consensus: AI is a serious tool, but it needs to be managed. "The system will tell you what it wants you to know, not what it needs you to know," one participant said. "You have to beta test, verify, and keep checking."
The more interesting thread was what brands are doing once they stop treating AI as a search engine and start building actual infrastructure with it.
One fast-casual brand described a full development pipeline they've built inside an AI-powered platform, loading in all their sales data from fifteen hundred locations to generate cannibalization rates, new store projections, and market opportunity analysis. It creates their LOIs, tracks franchise agreement timelines, flags who's coming up on deadlines, and feeds everything back into their main operations system. "It's a one-stop shop," they said. "We're just starting to get into it, but it's really changed how we operate."
Another participant — building tools largely on their own — described a utility monitoring portal that ingests invoices automatically, ties in weather API data, and flags anomalies in energy and recurring expenses like grease trap cleaning and hood cleaning. "I'm teaching myself to code through it," they said. "It's like learning a new language. Jump on the bandwagon or get left behind."
Lease abstraction came up repeatedly as a near-universal use case. Multiple participants described feeding contracts into AI tools and getting back in seconds what used to take a paralegal two or three days. "I need you to look at this lease, highlight midterm refresh and remodel requirements, flag anything unusual," one participant described prompting it. "Seconds. Literally what would have taken days." The caveat everyone agreed on: always verify the output.
One of the most useful frameworks in the room came from a participant who argued that the brands getting the most out of AI aren't throwing everything at it at once — they're building focused containers.
Contracting is one container. Menu management is another. Facilities is a third. Each one pulls in all the relevant data for that function, trains the tool on it specifically, and then starts generating real insights. "Unless you're driving toward a specific business goal, you get lost pretty quickly," he said. "The smaller the group training it right now, frankly, the better."
He compared it to a relationship. "Treat it like a six-year-old at first. Teach it what end of the screwdriver to use. When it gets to be a teenager, you have to slap it around a little. But eventually it comes back and says, 'You were smarter than I thought.'"
The existential question surfaced midway through: are dealmakers going to be replaced?
The room wasn't ready to say yes — but they weren't entirely comfortable saying no either. "You can look at anything on Google Maps, but actually seeing a site, feeling it, you can't do that from a computer," said one participant. Others pointed out that AI can now tell you household count, income level, daily per capita spend, and traffic patterns for any given block — information that used to require a helicopter ride with a real estate billionaire and a week of research. "It's Thompson Reuters on steroids," as one person put it.
The more nuanced take that got the most traction: dealmakers won't be replaced by AI, but they will be replaced by dealmakers who know how to use it. The ones who can run site analysis in a fraction of the time will have more bandwidth for the human work that actually closes deals.
Site selection sparked its own thread. Several participants are already pulling Placer AI data and credit card data into their models. The emerging view: the AI engine will increasingly do the analytical heavy lifting, but the source data is what matters. "If you have access to Placer AI's underlying data, you don't really need their value-add services anymore," one participant said. "The insight layer becomes commoditized. The data itself is what's valuable."
A few brands are experimenting with prompting AI to evaluate specific addresses — feeding it their performance benchmarks, trade area criteria, and brand parameters, then asking it to assess a location. "It's not quite there yet," admitted one participant. "But I need about a week to fine-tune the prompts. It's not that far away."
The closing question — where does this take us in the next two to five years — generated more honesty than predictions.
"I was hesitant to really get into it," admitted one participant, "but once you start using it, it's genuinely life-changing. Our jobs aren't going anywhere. This is a tool to enhance what we do, not replace it." Someone else pointed to the ATM rollout as a useful historical parallel: the same fears, the same conversations, a different outcome than anyone predicted.
The practical advice that kept coming back: start small, pick a specific use case, build the relationship with the tool before you trust it, and verify everything. "AI is not coming," one participant summarized. "It's here. You can either watch it pass you by or embrace it."
The moderator closed the session with what might be the most honest summary of where the industry is: "I feel like we're all just now getting really and heavily involved in it. I don't think we know yet what's really coming."
That part, at least, everyone agreed on.