Automotive News Daily Drive Dec 7, 2025

Dec. 7, 2025 | Bonus Episode: Steps dealers should take before picking AI vendors

Annotations will appear as you listen

0:00

12:31

Term

self-driving systems

Self‑driving systems let a car drive itself. They use cameras, radar, and computers to see the road and make decisions.

Brand

Rivian

Rivian makes electric cars that are good for off‑road trips and outdoor adventures. They’re known for their rugged trucks and SUVs.

Concept

software-defined architectures

Software‑defined architectures let car makers change how a vehicle works by updating its software, instead of redesigning the whole car.

Term

autonomy

Autonomy means how much a car can drive itself. Some cars need you to steer, while others can drive all the way.

Concept

mass market vehicle

A mass market vehicle is a regular car that many people can afford and buy. It’s not a special or limited edition.

LIVE

With nearly 5,000 specialists dedicated to supporting dealer financial services, Ally has the expertise to help you with your retail finance, F&I, and remarketing needs. We're all better off with an Ally. Contact your local Ally account executive to get started today. Ally do it right.

If you haven't heard the first part of that interview, go back and check out Thursday's episode.

You have them to obviously tryouts. And once you pick your vendor, small, medium, or large, what comes after that?

Well, I would, so the tryouts, I think are important before you just, you don't just pick a vendor and then like tryouts, I think you need to ask the vendor some questions.

So going back to data, because I think it's when you're selecting somebody, you know, just going from small, medium, large to, okay, now we're going to implement, you're again going to run into issues.

You want to ask some very important questions. For number one, if you do have a data infrastructure, so if you could answer the question, okay, this is where our data infrastructure is at, we have the resources internally, we have a good data lake blah, blah, blah.

Then you want to ask the vendor, are you able to connect to our data infrastructure? Do you integrate with our data lake? Or are you just connecting to the DMS or the CRM?

Because if that's the case, what are the steps you're going to take to then clean up, dedupe this data in order to make it hygiene data that can be used for AI?

That's the important question I would ask. In other question, I would ask the vendor is what model do you use? So when I say model, like I mentioned before, G-G-P-T, G-P-T-5 is a model.

Inthropic has a clawed sonnet 4.5, which model do you use? The reason, now if you're not technical, you're thinking, why am I, even if you're not technical, why would I ask this question? If I'm not technical, I don't know what any of these models mean anyway.

Well, it's more of a leading question, because the reason I would ask that question, let's say, I'm an engineer, I know what sonnet 4.5 is.

But even if they tell me, what am I going to do with that information? Well, it's a leading question into, so they go, okay, we use this model.

My next question would be, what if tomorrow, a new model comes out or a new version of a different company's model comes out, are you able to pivot to that one because it's better?

Okay, so it's not, in other words, it's not just about which model they're using that call the adapt is these things.

Yes, and the reason I ask that is, Mark, if you think about it, how often, if you read any news about AI, how often is there a new thing in AI, like in the actual tech AI space?

I'd say what every two months, there is a new release.

I was going to say less than that because you see updates all the time.

Exactly, and if a prospective partner completely ties themselves to that one model, and then tomorrow, a new one comes out that blows it out of the water, you are now automatically a deficit being partnered with them.

So then as you interview your potential vendors and tryouts, you're basically looking for flexibility and adaptability.

And honestly, honestly, internal testing against, do you test different models against each other?

Why did you pick this one? Why is it better than the other one? What did it do?

The other thing, and this happens a lot, is I would ask them is, do you track the activity that happens when we use your model for XYZ, whatever we decided we needed this for?

Now reason for that is AI, so the way I would describe generative AI today is it'll get you about 90% there.

So there is that 10%, maybe 95, depending on the use case, but it'll get you almost there, but there is a 5% to 10%.

And then 5% to 10% will happen. 100% of the time, it will happen.

In every implementation at some point, you are going to encounter something where this AI did something and you'll be like, that is not what it was supposed to do.

Now, in order for them to diagnose or fix that or see why it did that, they need something called event logs.

And all event logs are, is what was the AI doing when this thing happened? Can they track that?

If they cannot answer that question, that means when something goes wrong, you will go into the support line hell where they say we're working on it, and then nobody can actually figure out what's wrong with the problem.

So let's try to sum that up. So it's about picking a vendor who's flexible, adaptable, and somebody who is also accountable, who sees, who is aware that these things make errors and knows how to fix it within reason.

Exactly. That's the word that's really well. Now, the other thing you need to look internally too.

So you're talking to a vendor, you're seeing the problem you're going to solve.

I think a lot of dealers expect this and I think we've been in dealer world in general. We always run into this with software.

If you're a dealer principal, you're the tech person for the group, whatever you are, you get really excited at the demo.

Then they put it in and you go, okay, cool, I'll buy it. You sign, and then when implementation comes, it just kind of doesn't get used to nothing happens.

This can happen with AI as well. So internally, you need to be ready to change your process, you need to have a champion who will drive this.

I think the expectation is turnkey solutions, especially when it comes to AI, and I think a lot of times they're presented that way.

It's like, look, this thing will do everything by itself. It won't. You will need a champion to drive this forward.

A champion that works for the vendor or for the dealer for the dealer and internal champion who is a true believer in this new process and is willing to push it forward.

And when I say push, but does that have to be an IT person or somebody in the deal?

It can be somebody, it doesn't have to be. They don't have to be technical. That's not the point.

The point is, let's say we're putting a new AI that now will log all of our calls and make sure that everything gets followed up with.

Well, you need somebody internally to make sure, hey, are we using the system that now create to make sure people follow up?

Are people looking at that system? You don't need to be technical for that. Like that's like, you just need somebody to say, hey, we are changing this process.

This is what we're working through. You guys need to make sure you're looking at this and constantly communicating with the customers, like that kind of thing.

When we come back, more of our own Mark Holmer's conversation with Yuri Demidko of Fox Motors.

Rivian built its brand around Adventure, off-road escapes and beach camping, but CEO RJ Scorange sees a more robotic future built on self-driving systems as critical to gaining market share from slower moving rivals.

The next few years are going to look entirely different than the last few years, not just for Rivian, but I'd say for the automotive industry in general.

I'm Molly Boygon. On the next episode of Shift, part one of our exclusive interview with Rivian CEO, RJ Scorange.

He talks with our own Lawrence Eiliff about his vision for software-defined architectures, autonomy, and the launch of the company's first mass market vehicle in 2026.

That's available Sunday on the Automotive News Shift Podcast.

For over 100 years, Ally has helped dealers like you serve your customers by providing the best in-class products and services you need, and by remaining true to the automotive passion we share.

Your dedicated Ally account executive will work hand in hand with you to help you gain efficiencies, increase your product offerings, and work to improve per vehicle revenue.

Because we care about your business as much as you do, we're all better off with an Ally. Contact your local Ally account executive to get started today. Ally do it right.

Let's say that you've gone through this process, you've used a value-aided vendor that meets all of these benchmarks, and you have someone on staff to help bring this home. What happens next?

So the next thing is, and this one is, I've said this whatever step, but this one is very, very important. The next part is the structure of your agreement with the vendor.

So a lot of people, so now we have a lot of, again, I mentioned people are coming from the tech industry.

And in tech, the big thing is, you know, you have VCs funding them. So venture capitalist firms are funding them, that kind of stuff, and they need to hit targets.

And a lot of them are either trying to get acquired or they're trying to go public. Why am I giving you a finance lesson?

Well, this is really important. So a lot of these new AI firms, what they do is they come into our industry, they try it a couple of ways, and then they go wait a minute.

All these dealers jump on to things, and then jump off really quickly, dealers do us all the time.

And so then they go, well, we need ARR, annual recurring revenue.

In order to prove there is something in the pipe, and if we want to go public or we want to get acquired, we need to prove that there is money in the pipe that's not just hypothetical.

Like month to month contracts don't hit that.

So a lot of these vendors, the reason I tell you there is I'm giving a finance lesson, is what happens is a lot of them go, well, we have a minimum one month, one at 12 month contract.

Virtually all of them now say that minimum 12 month contract for the AI tool.

What you want to do is not argue over the 12 months, what you want to say is look, that's completely fine.

But if you're not like a single point in your group, what you say is, can we do a pilot with a couple of my stores to make sure it's working properly, and then with an out, and then we will commit to the 12 month contract.

So there is obviously the financial negotiation, but everybody knows that part.

What I'm saying, it's extremely important to do the pilot, because again, every single thing that we just talked about could go wrong.

Even if you said, okay, I think this vendor meets all this criteria, if something goes wrong, any one of these failure points will not waste your money.

And I think that's where a lot of dealers trip and get discouraged with this AI boom, is there like we put something in, it's not turnkey, nobody is using it, it's not, you know, it's messing stuff up constantly.

I need to get out of this and now it's a crap, and now I'm stuck there for 12 months too.

Yeah.

So in other words, then it's say you pick your vendor and you want to work with them, you negotiate that you try it out first on a pilot, because anything can happen.

And even the company itself is many of these are startups and something can go wrong with the company itself.

So you need out just in case that's what you say.

Yes, you need out for the pilot.

Now a good negotiating technique that I find is I usually say something along lines of look, if your stuff actually does what you say it does, after this pilot, I would be a complete idiot not to put it in all my stores.

If it's doing because I mean, if my problem is I want to increase online appointment percentages, just something simple.

And they go, hey, my tool will increase your online appointments percentages by 20% over the next 90 days.

Sweet, let's do a 90 day pilot if it increases by 20%. You solved my problem, why would I do it all my stores.

Thanks for listening to this bonus episode of daily drive. We'll be back on Monday with a brand new full episode of the show.