By Mark Hollmer

Yuriy Demidko warns dealerships shopping for artificial intelligence tools to be wary about boastful LinkedIn profiles.

“If it says something along the lines of ‘AI expert,’ ‘AI visionary,’ ‘driving AI forward,’ go ahead and disregard that company and proceed to the next one,” said Demidko, who helps choose AI tools as chief information officer at Fox Motors. “The salesperson selling you the AI is probably not the AI expert.”

Fox Motors, of Grand Rapids, Mich., ranks No. 44 on Automotive News’ list of the top 150 dealership groups based in the U.S., retailing 22,289 new vehicles in 2024.

Demidko’s advice is timely because, after some early stumbles, there seems to be more dealership consideration of AI tools.

A CDK Global survey of 251 U.S. dealerships from Sept. 15-28 found 63 percent of respondents were very likely or somewhat likely to invest in AI in the months ahead. And AI awareness is on the rise, with 51 percent either extremely or very familiar with artificial intelligence, up from 40 percent in 2024 and 41 percent in 2023.

“There are a lot of dealers out there looking for it,” said Leigh Ann Conver, CDK’s senior director of product marketing for data and intelligence.

Automotive News interviewed Demidko, technology experts and other dealership professionals who offered recommendations on how to effectively buy and implement AI.

Step 1: Decide what AI can improve for the dealership

“What problem are you trying to solve, or what are you trying to improve? That’s a very important question” to start with, Demidko said.

The answer should involve alleviating pain points, said Devin Daly, CEO of Impel, a retail software company that produces AI-based customer life cycle management software for dealerships.

Daly asked: “What is the problem you’re buying AI to solve? That you don’t want to deal with staff turnover? Is it that your server isn’t where you need it to be? Is it that you want to drive costs?”

It is important to consider AI’s merit as an employee tool, said Brian Benstock, dealer principal of White Plains Honda in New York.

“I think AI is a bandwidth machine,” said Benstock, who took an online course with colleagues through the Massachusetts Institute of Technology about AI basics for the workplace.

Dealerships also should factor in return on investment, Conver said.

“If they can focus on, what is the outcome? What is this going to do to help my business? That should be No. 1,” Conver said.

Step 2: Evaluate dealership data infrastructure

Dealerships should assess their data framework before they start shopping for AI tools, Demidko said, because that investment is wasted without a good “data lake” — all the information an organization has on its customers and business.

“AI sits at the top of data, and it’s essentially useless without data,” Demidko said. “It could be Einstein’s brain, but if there is nothing in there, it just doesn’t matter.”

Evaluating the data source is a messy but necessary step for dealerships, said Mo Zahabi, Cox Automotive’s assistant vice president of product consulting.

“Most dealers find their data is a mess,” Zahabi said. “It’s like trying to teach somebody to cook, then half the ingredients are expired or mislabeled. … If you have duplicates or a lot of bad addresses, it’s not really going to do any good.”

Dealerships with a solid data infrastructure should pepper vendors with questions about how their AI product would interact and how they would clean up the data to maximize that interaction, Demidko said.

Step 3: Shop for vendors of dealership AI tools

Shopping for AI is challenging because there are so many tech competitors, Conver said.

“Everyone has an AI solution,” she said. “It’s gotten easier in that dealers are more familiar with it. … The AI is getting smarter, but it’s also harder from a ‘which one is truly the right one for me’ perspective.”

Use platforms such as Google, ChatGPT or Perplexity to help cut through the marketing noise, Demidko said.

Factors he recommends considering when picking a vendor: customer ratings, intended use of potential AI tools and vendor size. Demidko said he prefers small- to medium-sized companies because “they’re the scrappiest and most innovative.”

Benstock and his team interviewed various AI vendors about their products before settling on testing three, giving them similar assignments. He was looking for a standout.

Step 4: Educate dealership employees, pilot the technology

After choosing AI technology and deciding how the dealership will use it, Zahabi recommends talking with employees about their new technology “co-worker” so that they won’t fear it.

“It’s like when they bring in a new employee, everybody hates that new employee because they think that employee is there to steal from them,” Zahabi said. “They believe that AI is kind of the same way. If you take that venom out up front … it’s going to be a lot easier for you to orchestrate the usage of it long term.”

Piloting should start small, said Benstock, who also is general manager of Paragon Honda and Paragon Acura in New York City. He signed off on an initial test that focused on using AI agents to handle inbound and outbound customer calls. Over time, his dealerships expanded the technology into inventory, incoming calls, emails and vehicle display pages.

It’s “crawl, walk, run,” he said.

Any goal with an AI installation should be to give a vendor partner every possible chance to succeed, Demidko said.

“Help the partner test this properly,” he said. “Make sure your people are on the calls and all of this gets done.”

Step 5: Train dealership employees on AI tools

Training can depend on the situation. If AI is used to handle Internet sales leads, for example, Demidko said training should remind sales employees to treat the technology as an assistant that helps their jobs.

Benstock implemented AI training gradually, closely coordinating with technology vendors. Usage training comes in tandem with making sure the technology works properly, he said.

“You have to constantly watch what’s happening because sometimes the AI [agents] can go out of bounds, and we want to make sure [it has a] limited language model to keep them on course and on track,” Benstock said.

A strategic AI road map and ongoing training are crucial pieces for any installation, Zahabi said. Dealerships also should designate someone internally as an “AI champion” who can optimize usage based on results, he said.

“Dealers that don’t have that … lack strategy, and they lack clear vision,” Zahabi said. “Generally, they’re going to be skeptical and uninformed about AI.”