In a scene from Ernest Hemingway’s 1926 novel The Sun Also Rises, Mike Campbell, a friend of the story’s protagonist, Jake Barnes, explains how he went bankrupt: “Gradually and then suddenly.”
The phrase has since been used to describe the pace of events ranging from technological advancement to societal change to personal growth. Software executive Nick Pericle recently referenced the scene before a group of HVACR professionals to illustrate how the evolution of artificial intelligence (AI) is about to accelerate, opening new opportunities for its use as a tool to train employees, boost sales, and generally help businesses run better.
“What we’re seeing has been happening gradually with generative AI, but we’re on the precipice of things happening suddenly,” said Pericle, a managing director at Profit Optics, a Virginia-based software consulting firm.
Pericle, who works with distributors and manufacturers, spoke in December to a standing-room-only crowd in Atlanta, Georgia, during the 2024 Heating, Air-conditioning & Refrigeration Distributors International (HARDI) conference. Offering several specific examples, Pericle said AI can be used to write scripts for sales calls, describe products based on images, review contract proposals, draft social media posts, glean action items from an article about inventory management, and much more.
One AI function, Pericle said, can even “listen” to a conversation, or to one person’s verbalized thoughts, and put the ideas into structured writing. That can be an easy way, he said, to take notes or create a formal list of standard procedures for running a business — just through speaking.
In one example, using the Open AI platform ChatGPT projected on a large screen, Pericle posed as a sales strategist at a mid-sized HVAC distributor and input information — a company profile, he called it — about his fictional business. He then fed the platform information about the company’s typical customers, and asked it to create “three distinct customer personas” in order to better understand them and learn what kind of messaging might work best and what products might best suit their needs.
“What it’s doing here is it’s going through its understanding (of) the context that I’ve given it,” Pericle said. “ChatGPT, generative AI models, are trained on the entirety of the internet, and so they have an understanding, likely, of your business — publicly available information — and just of the industry in general.”
Pericle then picked one of the ChatGPT personas — described as a “cost-conscious office building owner” — and asked for details.
“What this is going to do now is, it’s going to go back and in that same thread that I’m interacting with it in, it’s going to remember what I’ve already asked it, the detail that I’ve already given it, and it’s going to give me more information on persona one, the cost-conscious office building owner,” Pericle said. “If you’re within marketing or you’re within strategy, you know, you want to build out these personas so you have an understanding of who the people that you’re selling to are. It breaks this out.”
“I see this as a strategy guide or as a strategy partner for what you’re doing,” Pericle added.
Pericle also demonstrated the “vision” function of generative AI, inputting a schematic of a piece of HVAC equipment and the legend that accompanied the diagram and asking the program to tell him about it.
“So what the application is doing here is it’s going through, and, technically, it’s recognizing the image, it’s breaking it down into component parts, and it’s looking for patterns of information that it has previously been trained on,” he said.
Pericle at that point added a caveat: “Now remember, this is trained on the entirety of the internet, so you’re going to get some good, some bad, and a lot of garbage.”
But a distribution company that focused on training the platform with schematics of its products could find the vision application useful, he said, in converting that information into written form.
Pericle said he’s a big fan of ChatGPT and uses it every day; he also mentioned Perplexity AI and Claude, a platform from a company called Anthropic.
Perplexity, he said, is like Google, but its query results offer information with which users can interact. Claude, he said, can be used to turn work conversations, for example, into a written document about the way a business is run — something people at the business may talk about a lot but don’t necessarily write down.
“People understand their business better than anyone coming from outside does,” he said. “They don’t always have time to document, to build things out.”
One mistake that’s frequently made with AI, Pericle said, is treating it like a search engine: one and done. Generative AI, he said, can continue the conversation, building on what the user previously told it.
Pericle said AI “unleashes creativity” and will soon become a major business tool. He urged attendees to sign up for an AI service — paid subscriptions generally have more benefits than free versions — and start exploring it.
“My belief is that there will be organizations who are able to leverage these tools in such a way where they’re able to lower their operational costs and increase their output and efficiency,” Pericle said.