Todd Smith is the founder and CEO of QoreAI and former CEO of ActivEngage, the live chat provider he founded in 2007 and sold in 2018. At QoreAI, Smith and his team provide the “orchestration layer” dealers need to tie together disparate systems in an increasingly data-driven retail enterprise. He is the author of “The Intelligent Dealership: How AI and Data Transform Auto Retail.”

Shortly following an AI-packed 2026 NADA Show, Smith sat down with AI Dealer News to urge dealership decision-makers to push through the AI fatigue and be a leader in the adoption of new technology — or get out of the way.

AID: Todd, how did you get into AI?

Smith: I have been in the car business since 1990. I sold cars and then worked on tech startups, so I’ve been on both sides of the universe in our industry. I love data and data intelligence.

My interest in AI started back in 2015. I was invited to one of the first IBM Watson events, in Silicon Valley and San Fran. Obviously, the tech was nascent at the time. But it gave me a glimpse of what was coming. So I’ve just kept experimenting through time, watched GPs [genetic programming] come out and then full GPTs [generative pretrained transformers] take over.

But to me, one of the biggest changes happened at an Anthropic event in October 2024. I saw that I could send an AI agent to a bunch of websites to go pull data and fill out forms. That’s when I started writing my book. I knew someone would have to create the framework, to help dealers understand what is really coming at them, beyond just the fun of the AI that can do cool things.

AID: You have been prolific on LinkedIn. One theme I’ve noticed you return to is you can’t treat AI like SaaS. You should not be paying per user, per rooftop.

Smith: AI is built to pay for outcomes, right? You’re building a tool and you should be paid when it works. Our industry is a little bit of a laggard. For the last 26 years, you pay per seat or you pay per month. And dealers like that model.

AI doesn’t run on those same rails. And providers have compute costs and data storage costs that are incrementally going up. Ultimately, we will have to get to a model that is paid per outcome.

AID: Like a sales or service appointment?

Smith: Absolutely. And there is another way to look at it: Does the agentic agent save me time and energy, so now I can reallocate? I’m paying this person $30 an hour, and I’ve saved him five hours. That five hours is worth $150, right? So I definitely see a change in how dealers will pay for these tools in the future.

AID: You made that point in another post: Focus on replacing labor, not clicks.

Smith: We’re not going to replace employees with AI today. But what is an employee? An employee is a series of jobs, roles and tasks. Those tasks are going to be whittled out by AI. So I don’t see it replacing the job person yet. But if a job has 75 tasks, what happens if 10, 20 or 30 of them are now run by AI?

AI can’t do everything yet. But those tasks are going to go away, starting with highly redundant tasks, where I’m just inputting data into a system — and when you deal with AI, you don’t have that high human error rate.

AID: And employees will spend time programming and babysitting the AI, but in the hours they’ve freed up, they can do things only humans can do.

Smith: Humans will continue to do empathy work and strategy work. Tasks? Gone. Think about a sales guy. Do you think he likes, after a call, to log it in the CRM and put all the notes in?

AID: No.

Smith: AI will say, “You answered a call. It was Todd Smith. I’m now summarizing that conversation, then I’m going to look for next steps and push all that into the CRM.” Our roles are going to change. That’s all.

AID: You have said dealerships don’t have a data problem; they have a control problem. Was that in reference to the intelligence layer we’ve been hearing experts talk about?

Smith: We call it an orchestration layer, but it’s even more than that. If your data is not housed in something you own, then you’re not controlling the outputs. When my data bumps into your data, we have new data. That new data is called derivative data. And if that data is held not by you but your downstream vendor, they own it, and you have a control problem.

Let’s say you hire an AI tool. And it’s learning on your data, and it’s growing and getting smarter. And now you fire that vendor. All that learning just walked out. So control is going to be far more important in this agentic world. Because we’re gonna spend a lot of money training these tools and training these agentic agents. You don’t want them to walk out.

AID: What would you say to the dealer principal or GM who says, “Todd, I hear you and it sounds great. But this is all so new. I’ll sit on the sidelines and see how it all shakes out, and maybe I’ll invest in a couple years”?

Smith: Sell your store now.

AID: Wow! Just like that?

Smith: I’ve been in this business for a super long time. I’ve never seen a technology like AI, that changes so rapidly that it’s hard to keep up with. I tell every dealer, look, roll up your sleeves. Play with it. Fail with it. Have breakthroughs with it. There is no sideline anymore. The guys who actually dominate this, they’re not going to “little dominate.” They’re going to compoundly dominate.

Let’s say I sell 200 new cars and 100 used cars a month, and I’m servicing about 2,000 cars a month. Now I’m going to start using AI to make my company more efficient, because efficiency is the last horizon. If I can be 3% more profitable — then 6%, 12%, 18% — I can start discounting on the front. I’ll margin you to zero. Then I’ll margin you to negative. I’ve built a far more efficient operation. You can’t beat me.

So I would tell that dealer or GM, if you don’t want to play this game, then don’t be in the game. Retire, be happy, go do something else. But if you really want to own a car dealership in this world, you play the game. And the objective is to get really, really good at leveraging this technology.

Tariq Kamal is the publisher of AI Dealer News as well as Auto Dealer News and Dealer Agent News.