Lessons From the AI Frontlines in a BHPH Dealership

By Tiger Okeley, Executive Board Member, Oak Motors/Indiana Finance Company

The artificial intelligence (AI) wave has crashed over all of us in the last few years, and many auto dealerships are struggling to keep their heads above water. If you're anything like me, you've probably seen trade magazine coverage of AI vendors that make big promises about their capabilities. Or you've heard stories about the tools major industry players have already deployed. It's enough to make your head spin.

While the buzz is tough to ignore, bridging the gap between AI's promise and the reality of implementing these tools in a working auto dealership is a significant undertaking. Most organizations lack the resources to hire consultants or experiment with AI solutions that don't add value. As the operator of an Indiana-based Buy Here, Pay Here dealership, I've faced those questions too. However, I've decided that despite these early challenges, we can't let this technology pass us by.

I've attempted to deploy AI solutions in my organization and pivoted when they didn't perform as expected. Here's what I've learned during that process and how these lessons can help other BHPH operators start their AI adoption journey.

Visions for AI Adoption in a Dealership

Our organization has a pretty clear vision for what we hope AI can do for us. Ideally, AI won’t replace our team because we're in a people business. We're not selling widgets. We're helping our customers make impactful financial decisions that are mutually beneficial for both parties. In our opinion, people will always need to take care of other people.

However, like many organizations, our work involves repetitive and time-consuming tasks, stubborn redundancies and communication barriers that slow down our workflows. Those roadblocks exist between our internal systems and departments, but also between our customers and employees. If we can deploy a consistent, appropriately skilled AI to take on some of those tasks, we'll be able to help our customers much more efficiently.

To accomplish this vision, we need an AI solution that can seamlessly access data across departments like finance, sales and service, while creating meaningful outputs our teams can use. We have a robust IT department, with expertise across hardware, programming, security and compliance. That’s not common in most dealerships, and it’s an asset for us as we navigate AI implementation. While they feel hopeful, they believe a deployable AI system that accomplishes everything we need is still a couple of years away. But that hasn't stopped us from experimenting with AI tools in individual departments. Our first major experiment taught us some valuable lessons about the current state of AI technology.

Learning Through Action: Our Failed Experiment

About a year ago, we contracted with a company to deploy an AI chatbot in our collections department. The goal? To automate compliance requirement tasks that our team must accomplish with every engagement, primarily involving a defined verification process, such as confirming identities and validating account information, to ensure we're talking to the correct person. While these tasks are critical, they don’t directly add value to the process.

We spent months researching our options, choosing a vendor and planning our deployment. We also spent weeks training the chatbot on our data. But on launch day, we killed the project after only a few hours because it couldn't handle anything beyond sending an initial greeting.

If a chatbot couldn't accomplish basic verification tasks in a controlled environment, it wasn't ready for the complex, multi-step processes our industry requires. Its failure also made it clear to our team that current AI solutions aren't ready to function effectively across complex departments and data sets.

This poor result doesn't mean that we've given up on AI adoption at Oak Motors. Quite the contrary. What it did do was clarify our requirements for AI tools and help us establish a few guidelines for adopting them in the future.

Our Business is Complex and Very Human

This failure reinforced a fundamental challenge in the BHPH industry: even our simplest processes involve complex, interconnected customer relationships that AI can't yet navigate. Imagine a customer who needs help with a significant vehicle repair problem. This customer has never missed a payment and never brought the vehicle in for service. They've also put 140,000 miles on the car in two years — twice what a typical customer would drive. How would you help this customer? This isn't a collection problem. It's also not a service problem. It's a customer lifestyle issue, which is a much more nuanced topic.

That level of nuance and context simply isn’t something current AI tools can process. In a year or two, it may be able to bring all this information together in these complex situations and present the variables to one of our team members. Ultimately, a human will need to review and approve the recommendations and discuss the options with the customer.

Hopefully, AI will someday have the capability to collect relevant information and then make consistent, accurate and unbiased decisions that are free from subjective factors like not getting a good night's sleep or feeling upset from a previous customer interaction. Those capabilities seem a long way off. While we wait for AI to mature, we're taking a more strategic approach to implementation.

The Next Steps of the AI Journey

Our team is back at the AI drawing board. This time, we've selected a different vendor that offers a new product for our customer retention and sales operations. Before choosing this vendor, we saw a sample deployment that was much better than the chatbot we tried to implement earlier. We're currently training our AI bot to handle the easy stuff, and the results are promising so far.

As we continue our AI adoption process, we're going to hold fast to a few core principles.

• First, we'll take small bites out of the apple from both a budgetary standpoint and our time requirements. These projects can easily become a distraction from our core business, so it's essential to stay focused on what matters.
• Second, we're going to enter every project with a clear goal of what we're trying to accomplish. We won't deploy a tool just because we can. Instead, we'll make sure we see meaningful results that add value to our business.
• Finally, we need to see proof of concept before investing our resources. If a vendor can't demonstrate results upfront, we're not interested.

Along the way, we're continuing to build strong AI literacy in our organization so we'll be primed to take advantage of future opportunities. This measured approach ensures we stay competitive while building the foundation for future AI success.

Looking Toward the Future

The AI landscape in our industry is evolving rapidly, and even though the technology isn’t perfect yet, dealers who start experimenting now will be in the best position later. We've had setbacks and learned a few costly lessons. However, they've made us much smarter about our goals and how we evaluate potential new tools.

The reality is that our larger competitors are already using their resources to implement AI solutions, whether we're participating or not. My advice to fellow dealership owners is to get involved, but stay smart. Take measured risks, demand proof of value from vendors and don't be afraid to pull the plug when something isn't working. The cost of trying and failing today will be much less than the cost of being left behind tomorrow.

Dealers who can embrace change without sacrificing the fundamentals of customer service will thrive no matter what the future brings. AI won't replace our ability to help our customers. But, it will empower us to be just as responsive and supportive as our customers expect. That's a future worth working toward.