From Sci-Fi to Reality: Discovering and Harnessing the Power of AI

After existing primarily in science fiction stories for decades, artificial intelligence is having a real-world moment. Experts say AI is here to stay and that it will transform franchising in significant ways.

Ryan Farris, president and COO of AlphaGraphics, PostNet, Multicopy, World Options, Pack & Send, and Mail Box Etc., has been charting the AI course for the companies he leads.

"If you look at the Goldman Sachs report, they show four in 10 jobs can be 30% more productive using generative AI," Farris says. "When we start with the mindset that maybe my individual teams could be 30% more productive, we want to introduce those colleagues to AI. Then we want to introduce them to the guardrails of how to be safe with the data."

Franchisors use AI for market analysis, research, training, support, sales, marketing, and more. Tools are being developed to operate drive-thru windows. Chatbots will soon be able to tell what potentially life-saving lab tests a person needs. Someday, AI might be in charge of scheduling your child's swim lessons and playdates.

Business leaders owe it to themselves to investigate the possibilities. There will be pushback, but Kelly Crompvoets, vice president of marketing for Any Lab Test Now, says technological advances have always generated skepticism.

"You switch everyone over to a new CRM, and that's hard to change," Crompvoets says. "I used to train people. And I used to say to my classroom of franchise owners, 'Look around. Nobody here has a haircut, clothes, or whatever from 1975."

For years, AI was a sci-fi writer's dream. Now, it's a tool available to anyone with an Internet connection, which leads to a natural question: "How will AI be used?" These are early days, and the answers are wide open.

The four workstreams

Farris says people often imagine sophisticated AI solutions that require massive investments and extensive overhauls, but AI can start small and inexpensively. "Make it easy. Start simple. Crawl before you run," he says. "You might find that your team, once they understand how to use it in their own roles, departments, and functions, will then give you better ideas of how to invest for those bigger initiatives."

He and his team developed four AI workstreams. It starts by getting people used to working with it to conduct research, build tables of content, do translations, perform competitive analysis, and discover their own uses. The goal is to increase efficiency without making AI seem foreign or threatening. That involves letting staff members play around and have fun with accounts that are shared by departments.

"They can see each other's scripts. They can see what everyone was doing to play with it," he says, "and then they talk."

The second workstream is about streamlining utilities. For a time, different people brought recording tools to each meeting, but only one recording was needed. A team looked into each tool and created a standardized list of approved AI as part of a companywide policy.

"This is what every employee is going to be trained on. This is what they're going to use. This is how they're going to use it. And this is how we're going to accelerate best practices," Farris says, "so we're not individually learning, individually reinventing the wheel. Now, it's an organizational effort."

The investment increases for the third and fourth workstreams. The third involves giving AI access to company data to improve efficiency. Reports that might have taken the finance team days to compile could be ready in seconds, allowing for on-the-fly financial analysis and trend comparisons.

"The fourth workstream is sales and marketing," Farris says. "What can we do to actually generate more sales and grow our network better, faster?"

Examples include automating customer messaging, prospect identification, and social media posts. Everything needs to be personalized, and it needs to be clear and correct, so it doesn't give a prospect a reason to shut the proverbial door.

Farris cautioned that workstreams three and four should be treated as aspirational in the beginning. Starting with simple tasks and scaling up allows for better adaptation and strategic investments. "Everybody seems to think you need these big, huge, lofty enterprise solutions," he says, "and they're missing the fundamentals."

Speed, speed, speed

At Any Lab Test Now, the team is working with a vendor to create an AI-driven chatbot designed to guide customers to appropriate lab tests. Crompvoets says the chatbot will ask the consumer questions about their situation and symptoms before providing tailored recommendations. If a 35-year-old woman is considering pregnancy, the chatbot will suggest relevant tests to help her start the journey. Crompvoets says the goal is to reduce the time she spends searching the website and give her quick, accurate suggestions.

"AI is all about more data, more learning, becoming smarter, becoming refined with what it returns back to the consumer," Crompvoets says, "but we're using it so that it can be a much more conversational approach."

To protect a client's privacy, the AI chatbot will not collect any personal information, such as names or contact details, during the initial interaction. "If they then choose to give us their email address a little bit further down the funnel, then at that point, we can start to market to them," Crompvoets says. "But the first interaction is really just trying to say how our company can serve them."

The project, which began at the start of the year, is in the testing phase with a planned full deployment in Q3 and a system-wide launch scheduled for the company's September conference.

That's also when another AI project is expected to debut. It's a platform to aid franchise owners by automatically indexing and scanning documents, eliminating the need for manual tagging. "Franchisors send documentation. They send written updates. Everything is documented in the library," Crompvoets says. "Somebody might have gone looking for something traditionally online. They get frustrated. They can't find it. They're searching for a different name than what somebody tagged it or titled it. They get frustrated. They call their business coach, who may or may not be able to answer the phone or be able to answer their email."

The AI is being designed to serve as both library and librarian, streamlining access to documentation. Crompvoets says the approach mirrors the in-store experience, emphasizing quick and efficient service to meet consumer needs.

The company also uses AI internally to help with brainstorming, writing the first drafts of emails, and coming up with new promotions. "Sometimes, you need time to be thoughtful and to let things marinate, and that's where AI has definitely helped. You can just kind of spit out your unedited brain dump and see what it comes back to you with. That's been super helpful," she says. "It's been helpful for the team. We have one person on the team who looks to AI as almost like her assistant or a co-worker to bounce ideas off of."

The plan is for AI to become an integral part of business operations. Crompvoets expects some franchisees will resist the change at first, but new owners will accept AI as part of the way Any Lab Test Now does business.

"I just hope people see the importance of AI for their businesses," she says. "I hope they're not as slow on the draw with AI as they may have been with other technology in the past."

Dive into the future

Dennis Leskowski took over as chief technology officer at Goldfish Swim School less than a year ago. He brought a history of software development and AI experience. "What I feel like is that we're at the beginning of the space race in the 1960s. There is a moon out there. There's a Mars out there," Leskowski says. "The question becomes less about how are we going to get there and more about where do you want to go?"

At its core, AI is about data, he says. It comes into play anytime there's a need to think, analyze, recognize trends, or match data points. For instance, scheduling is matching someone's availability (A) to open slots (B).

"My wife and I were trying to get a playdate for our daughter to hang out with another person's daughter," he recalls. "I said, 'Listen, in about a year and a half, it's going to be your AI talking to her AI. They'll both know the schedules, they'll both know the preferences, and they'll just set it up. That's what's going to happen.'"

Leskowski says that for AI to be truly effective, two main prerequisites must be met: data organization and a structured software development lifecycle.

Without well-structured data, AI systems can't function properly. He says the technology team at Goldfish serves as educators for the AI, ensuring it learns the right subjects and corrects any errors, much like a teacher would with a student.

"The formal term would be metadata tags," he says. "If you just threw out a bunch of data, a bunch of numbers, and it doesn't know, 'Oh, that's revenue. Oh, that's enrollment. Oh, that's my workforce employee.' Then it's just a bunch of worthless numbers."

The second prerequisite is to apply the software development lifecycle to AI. This means understanding the entire process from idea inception through development and testing to deployment and ongoing refinement. "AI is not a switch where it's turned on and turned off," he says.

He compared working with AI to walking into a gold mine without a bucket. AI's potential is exponentially larger than anyone's ability to fully exploit it. He and his team prioritize projects, develop them, and test them.

With access to data from a quarter million swimmers, AI helps identify patterns and trends that humans might miss, assisting the curriculum team in refining the lesson experience. AI also helps analyze seasonal sales data, allowing the company to focus on precise areas for improvement. "Instead of taking a sledgehammer and trying to fix everything, it's using a scalpel," he says.

Goldfish Swim School has chatbots and other AI features under development. Some are being tested on the company's website, a few are being tested by franchisees, and others have marketing and curriculum consultants putting them through their paces.

Leskowski says patience is key; there's no reason to rush deployment. If someone asks for the hours of operation, and the AI responds with "hamburger," it wouldn't reflect well on Goldfish Swim School.

"Do you remember when those voicemail prompts came out like 20 years ago?" he says. "And everyone's getting caught in a loop of press one for this, press two for that. Everyone started pressing zero because they just wanted to talk to a live person."

Since Goldfish Swim School deals with children's data, privacy is paramount. While aggregated data for trend analysis poses little risk, using AI for personalized communication requires careful consideration.

"I can envision a world in which an AI chatbot is interacting with a person, and we know things about them--who they are, where they live, which children they are--because we need to know the ages of their children," he says. "But I'm hesitant to apply AI to that simply because we're talking about a children's service industry."

Ultimately, AI can handle numerous mundane tasks, but important decisions, including how to protect customer data, should be made by people. "What can't be handled by AI is what we call the 'everyday golden experience,' our edge, as well as the swim lessons themselves," he says. "We joke that we'll never have robot swimmers helping kids in the water."

Drive-thru AI

Bojangles, which specializes in chicken and biscuits, has an AI taking drive-thru orders at more than 150 of its restaurants. "Bo-Linda" is a conversational cloud-based AI platform, and it can take guest orders more than 96% of the time without help from human staff members, according to Richard Del Valle, chief information officer.

"I wouldn't be surprised if, by the end of the year, we were in the mid-200s in terms of deployment, maybe even higher," he says.

When Del Valle joined the company, Bo-Linda was on the back burner, but it had been in listening mode, recording transactions, learning how people order, and connecting words to actions. "We kicked Bo-Linda into high gear," Del Valle says. "We recognized very quickly that it was something that needed to be prioritized for us."

The AI had plenty of learning to do before it was ready for the public. For Bo-Linda to function, the Bojangles team and the vendor had to make sure all menu items, including multiple variations, had a corresponding number that the AI could recognize. Those numbers have to be consistent across the network, or it won't work.

"Occasionally, we'll get a guest who wants a hollowed-out biscuit. They want their sandwich, but they don't want all the fluffiness of the biscuit in the middle. They really just want the outer shell, right?" he says. "Well, for 30 years, we handled that by typing a message into the POS called 'hollow biscuit.' But when you launch a product like Bo-Linda, you can't do that because Bo-Linda needs a menu item to reference."

It was a difficult, detail-oriented job to provide the information for Bo-Linda, but now, all the company and franchise restaurants use the same item numbers for each product, which made the rollout possible.

Bojangles also uses AI to improve its food forecasting and labor scheduling. An AI engine can analyze sales data from previous years to predict how much food will be needed on a given day. This helps prevent running out of popular items or having too much waste.

"It's like, 'Hey, Mr. General Manager, here's your food order based on your sales forecast and what you have on hand. Here's what you need,'" Del Valle says. "The manager can focus on other things."

The AI engine can also create labor schedules. By considering sales data and employee skill sets, the AI can suggest a schedule that will ensure that there are enough staff members on hand to meet customer demand.

"These are very powerful tools. We like them because they improve the quality of life of our managers, and they improve the guest experience," Del Valle says. "If I can get the right people on shift at the right time to deliver great service, the guest is going to get great service."

AI is there to augment humans, not replace them. "These tools provide guidance and suggestions. We tell our managers all the time that the most important AI engine we have is the one between their two ears," Del Valle says. "The tool doesn't know everything. It might not know that you have a parade scheduled for next week. Or it might not know that you had a parade last week."

It was decided that Bo-Linda shouldn't sound human, so customers wouldn't think they were getting fooled. Even so, Del Valle says Bojangles' employees often think of Bo-Linda as a member of the team.

"As human beings, we create this relationship," he says. "Nobody knows this technology better than I do, and I thank Bo-Linda all the time, and then I think to myself, 'Why did you just do that?'"

Caveats

As AI becomes more integrated into our daily lives, understanding its strengths and limitations is crucial. AI requires a certain finesse to be used effectively. The ability to ask the right questions and to interpret its responses is more of an art than a science. For AI to function correctly, it needs human beings and their experiences, says Mike Chadwick, executive vice president of growth with Ironside, a data science and management company that helps businesses get started with AI.

"The human brain is absolutely freaking amazing, and the number of data points your brain takes in, processes, and understands what to use and what not to use is unbelievable," Chadwick says. "AI still is way behind. It can take amazing amounts of data, but it doesn't always understand the nuance of the context and what to ignore and what's important."

An easy introduction to working with AI is to ask it to write a business email. It's great at producing a first draft, but don't click "send" yet. Chadwick recently received an email from a client. The email included a note from the AI that advised the sender to edit the email before sending it. "You always have to edit," Chadwick says. "You have to add your special sauce to it and your insights."

While AI can handle structured tasks within a defined set of rules, it falters when faced with the unexpected. For example, ordering a custom item at a fast-food restaurant could be a straightforward task for a human but might confound an AI.

"It needs to be kept targeted and specific," Chadwick says. "If you're in a certain game board--that's a metaphor, right?--there are rules and boundaries around what this is going to do. AI will be phenomenal. The more that board game grows and the more complicated multiple factors become, the more it's going to struggle because there's going to be more ambiguity in the answer, and AI doesn't like ambiguity."

Humans provide context and make decisions based on a variety of factors. AI, on the other hand, can process vast amounts of data but often misses the subtle cues and contextual nuances that guide human decision-making.

"The data will come from literally almost an infinite number of sources, and some of those social sources will be so minuscule in their effect that you can ignore them," Chadwick says. "AI can't really ignore anything. It has to go through all the permutations, and it doesn't always know which one of those is more important at a given moment."

As AI continues to evolve, the key will be finding a balance between leveraging its computational power and retaining the human touch that provides context and nuance, Chadwick says. A decision to expand a business might have 1,000 data points in its favor and only one data point against it, so an AI might say, "Go for it." But what if the plan is to expand operations to Israel on the day after Oct. 7, 2023?

Chadwick says AI should be treated as a tool, and human operators and/or team members should follow time-honored advice: "Trust but verify." The art of asking the right questions and interpreting AI's responses is where the true power lies, ensuring that technology serves as an effective partner rather than a flawed replacement.

"Getting all that free-form data and an answer to those things is just unbelievable power for someone who has knowledge and understands the key factors in play," Chadwick says. "The human being is bringing the context to the conversation. They're able to make gut-feel decisions based on their experience."

AI Advice

If you're planning to introduce, or continue introducing, artificial intelligence to your organization, here are some tips for getting started:

• Start small. Introduce AI tools that are accessible and easy to use, allowing individuals to explore their potential.
• Encourage discussion. Give employees ways to bounce ideas off each other and share what they learn.
• Formalize within the organization. Develop a structured approach with small budgets initially and expand as the organization becomes more comfortable with AI.
• Standardize utilities. Avoid using multiple similar tools and streamline processes with standardized utilities.
• Offer comprehensive training: Ensure employees receive thorough training and support to maximize the benefits of AI tools.
• Find departmental champions. Appoint AI champions within each department to lead initiatives and facilitate learning.
• Do regular follow-ups. Conduct 30 and 60-day checks to evaluate progress, address challenges, and ensure continuous improvement.