They’re talking about AI that actually makes a difference in day-to-day dealership work. Not just something cool—something that helps sell cars or reduce problems.
The “human element” is the idea that AI performance in a dealership depends on people guiding it, validating outputs, and integrating it into processes. The speaker argues the differentiator isn’t the AI itself, but vendor focus on human workflows and pain points.
“Internal friction” describes added workload, process changes, or coordination problems that occur when new tools are introduced without fully considering operational impact. The speaker uses the website rollout as an example: it created new responsibilities and management needs.
This highlights a common dealership tech pitfall: implementing a tool without accounting for ongoing operational ownership. Even if the tool is beneficial, someone must manage content, updates, and performance to keep it effective.
They’re saying the goal of using AI isn’t just to do tasks—it should help the dealership sell more vehicles. And you should be able to measure that improvement.
“Smoke and mirrors” is a metaphor for something that looks impressive but doesn’t produce real, measurable outcomes. Here, it’s used to criticize AI deployments that lack post-launch reporting and evidence of impact.
They’re saying you need good information in place before you turn on the AI. And you also need to check results afterward to make sure it actually helped.
A marketing funnel describes the stages from attracting shoppers to converting them into buyers. The speaker connects AI-enabled campaigns to filling the funnel and then turning that demand into actual car sales.
A “historical trail of data” refers to having enough past performance records to validate what the AI will do and how it should be used. Without that baseline, it’s harder to prove outcomes or tune the system to dealership-specific realities.
ROI is basically “did this cost more than it helped?” You look at the results and compare them to the money you spent. If the numbers show a benefit, it’s worth keeping.
It means figuring out what’s changing and why, then making smarter decisions based on that. The idea is to use AI to spot patterns and guide what you do next.
They’re basically saying humans bring the real-world understanding that AI lacks. Staff know the dealership rules and customer context, so they help AI make better decisions.
Use cases are the specific jobs you want the AI to do. The episode says AI works best when it’s matched to the exact tasks your dealership handles every day.
OpenAI is a well-known company that makes AI tools. The episode is saying that even with top AI brands, you can’t just plug it in and assume it will handle everything perfectly.
Anthropic is another big AI company. The takeaway is that even strong AI tools need the dealership’s guidance and supervision to work well in real sales and service situations.
Gamification is when a tool uses “game” tricks to keep you interested. In this case, it’s like the software keeps nudging you to come back and keep working.
“Blanket AI” implies a one-size-fits-all AI solution that isn’t tailored to the dealership’s specific workflows and needs. The speaker argues that this kind of generic AI is often more about attention/hype than solving the real operational problem.
Stream Companies is the company talking about how AI should be used in car dealerships. They’re saying they help dealers use AI safely, not just “turn it on and hope.”
They’re talking about the ways AI can go wrong in real life. Their point is that dealers need help and guidance so the technology doesn’t create new problems.
They’re saying the company doesn’t just hand over software. They also teach dealers how to use it and help them plan how it fits into their day-to-day business.
If the information going into the AI is wrong, the AI’s advice will also be wrong. That can cause the dealership to chase the wrong leads or make bad decisions.
It’s when you look at numbers without the real context around them. In a dealership, that can lead to advice that sounds smart but doesn’t fit reality.
It means you’re only paying attention to one thing and ignoring everything else. In business terms, that can lead to bad conclusions from partial information.
A holistic view means you look at all the information together, not just one number. That helps you make a better decision.
Concept
zoomed in vs zoom out
The “zoomed in vs zoom out” framing describes how focusing on a single indicator can cause you to miss other important signals. In dealership operations, this maps to balancing one metric (like a warning) against the overall customer and sales-floor picture.
LIVE
All right, gang, welcome back to this episode of the Dealer Playbook podcast.
I have my friend.
People don't know this, that we've had lots of conversations.
My friend Mackenzie Wiltrout from Stream Companies.
I'm so glad you're here.
Thanks for joining me on the Dealer Playbook.
Michael, thank you so much for having me.
This is really an honor to be on your podcast.
Oh, shucks.
Now I'm blushing.
What is my first NADA?
I want to say that.
So is it really?
Yeah. How's it going?
This has been so exciting.
I've been having a great time.
This has been amazing energy here this year.
And over at our booth, we've had a lot of fun stuff that we're unveiling.
So really just a banner couple of days.
I just got to say, by the way, Stream always has some of the best swag at NADA.
I'm just going to say it.
Stream and its affiliates all have some of the best swag,
which I had to run to registration and pick up one of the exhibitor backpacks
because tomorrow, tomorrow's shopping day, friends.
OK, I'm so excited to get into this.
You have such a tremendous access to data, to innovation,
to so many different things that you guys are building and working on,
things that you've seen work.
So I'm curious with the market flooding and the NADA flooding
with new AI tools in 2026, what does it truly mean to find impactful AI?
This is a great question because if you've walked the floor here at all
and I know that you have every other booth, maybe even every booth
has AI somewhere listed.
So there is a flood of AI tools out there right now.
But when it comes to the impactful AI, I think what sets the tool apart
or the vendor apart is not even the tool itself,
but still their focus on the human element of it and focusing on people
and their pain points and problems, how they're actually going to use it
to solve problems.
And that is a question that I have heard at our booth and I've heard it
just kind of floating around, which is I was just actually going to solve a problem.
Yeah, I get that the AI knows that there are issues out there
and it has access to data and it can point, can point at things.
But we need a little bit more than pointing in the year 2026
for dealers to move metal off the lots with AI in their arsenal.
What you're making me think of is, you know, back in the day when we introduced
websites, then all of a sudden we it was pitched as a solution,
but it actually did create some internal friction because now we're like,
well, now we need somebody to manage this website.
Now we need and look at how that's evolved.
But now we need teams to manage websites.
Now we need a whole department.
Now we need and my concern is to your point, and I love this impactful
actually solving for something.
How do we navigate that to make sure that we are actually solving a problem
and not just creating a new one?
Oh, boy, I have fielded that question quite a bit.
OK, and I always say the proof is in the pudding.
Yeah. So you bring on a new tool.
You're like, I've got this new tool in my belt here in my my tech stack.
I'm really excited.
The AI, everybody's telling me it's going to solve all my problems.
You know, it's going to help me sell more cars, service more vehicles.
It's going to do my taxes.
It's probably going to solve all my relationship issues, all of these things.
Right. But at the end of the day, if you have the tool and then there is
no reporting data after the fact that is actually showing you the tool
made a difference, there is no case study, then we have smoke and mirrors.
So we always prize the data ahead of time, the data that underpins
the AI that comes before the AI.
But I'm thinking about what comes after the flood and the data after the fact
that shows that, OK, so this dealer was using this to build strategy,
build campaigns to fill out their marketing funnel and then to actually sell
cars within their market.
We need that reporting after the fact that shows that you have your return on
investment. And to be honest, because we're still sort of at the that genesis point,
right, these AI are still so young, you know, they're like they're like shiny
little babies, right? Yeah, still so young, which is odd to say about a machine,
but it's a fact. We're still so early on in understanding what AI
can do for us that, to be honest, we don't have that historical trail of data
quite yet. But I think if you are really, really educated and you're working
with a vendor selling you the AI that is also educating you and taking
the time to teach you how to use it so you really understand and to your point,
you don't need somebody on staff to manage it.
You should be able to see a return on investment and there should be numbers.
There should be proof in that pudding that shows that this made a difference
in your strategy from now to the before times before you had the AI.
This is moving beyond technology in a vacuum.
You know what it reminds me of?
It makes me think of
getting blood work done by your doctor
because you could work out and eat healthy.
You could do all of the things, but it's the trailing data of three months
history of blood work that's going to help you understand that full,
that that loop back to the beginning that will then inform the things you do
next to your between that now and your next blood test.
Correct.
So how do we how do we encourage that shift in a way that is
palpable and not like a huge change management nightmare?
How do we move beyond technology and start diagnosing the shifts that we should make?
Again, I think it comes back to having people in the loop,
in the process, working with you to understand what the AI is doing.
I feel that this something I've seen quite a bit is
we're telling people, not us.
Other vendors are saying the AI is going to solve everything.
So they just put it into your hands under the misconception that it is self driven.
It'll be able to answer your questions.
It'll last so the sun, moon and stars for you.
Exactly. Like, oh, it's it's an intelligence.
So it's smart.
It can tell you how you're supposed to use it.
Right.
That is such a misconception.
You're saying AI needs regular eye.
Yes, yes.
A artificial intelligence needs organic intelligence in order to drive it
correctly and to understand your use cases.
And that's another thing I see so much like
general purpose, one size fits all AI out there.
And I'm not throwing shade at any of the big enterprise.
I'll throw the shade.
You know, our open AIs of the world or the anthropics.
But you get your your general LLM, your standard Chapa garden variety,
which is what is mostly out there right now and what people mostly know or think
of when they think AI, they are smart.
They can do a lot.
They're capable to an extent, but it's general purpose.
They are trained on all the data.
And when they have all the data to look at, they often don't know
for this particular use case or this particular problem, I should actually be
looking at that data set or I need to maybe look over here.
So and a worry of mine is to your point about these, I love the garden
variety analogy or phrase.
Because my worry with them is that they just affirm my bias in a vacuum.
They're not challenging my thought processes.
They're actually, I feel like they're seeking to understand my bias
just so they can affirm it.
And that's the context that they work with.
Well, Michael, to quote chat, GBT, you are absolutely right.
They are the most the most obliging pieces of technology you could possibly have.
You bring up a great point, but Kenzie.
Oh, my gosh, that's so.
So speaks out both sides of its robotic mouth.
And that's a fact.
Right. Yeah, because I've been in threads that I've even tried programming,
right, that when I'm like, I feel like I've given it all the information
in the world that it needs to contextually help me.
And it still just affirms my bias.
And then sometimes I'm like, no, that's not what I'm thinking about.
And they're like, you know what, sorry for making that mistake.
And then it just outputs the same crap anyways.
Yes, because you have to remember that these are enterprise products.
These companies are trying to make profit off of them.
And so there is a sort of gamification of it.
Like, I don't know if you've ever gotten into video games or anything like that.
But, you know, in the game world, systems are designed to make you want to keep playing.
There's a certain grind involved where it's always dangling a carrot to say,
keep playing, keep investing time in this.
Perhaps there are micro transactions involved.
And so keep spending money.
AI being so obliging and wanting to just sort of agree with you is the same exact thing.
It's a carrot that is dangling saying, keep using me, keep using me.
You need me, especially if if you're offloading some of your cognitive ability
onto the AI and it's it's taking the place of things that you used to do yourself.
Right. This is this is interesting, too.
Because to that point, we've got a comment here on our live stream Landon says,
can't tell you how many times I've had to handle all of that on my own
because the owners didn't think it was important to invest in technology
or having actual teams of people to handle those areas.
And you're saying, well, yeah, because we don't even have a full understanding
yet to be able to really get to the heart of our needs.
Because just a blanket AI is not actually the need.
It's it's a hype piece meant to get your attention.
It is. And again, it's dangling the carrot.
It just wants you to use it to talk to it.
You know, not that it's sentient or anything like that.
Maybe there's some disagreement about that online.
But, you know, yeah.
Well, when that time comes, I will tell you right now,
I will not be producing episodes of the playbook anymore.
I'm going to be digging a hole in the ground and going to live for it.
So but this actually brings brings up something I want to ask you
as we as we wind into our closing.
What is Stream's vision for helping dealers avoid these technological pitfalls
while still taking advantage of what AI can do well?
The end of the day, yeah, we are partners to our dealer clients.
And that is a partner, whether you're somebody who's managing an account
in trying to drive strategy or we're providing technology.
We're a partner in the technology, too.
So we're creating this AI that is very purpose driven
within our orange OS system that is launching in 2026.
And yes, it was built for dealers by people that understand dealers.
But this is not going to be a scenario where we just say,
have at it, go into the dashboard, play with the AI, let it solve your problems.
You know, we wash our hands of it.
Never, never.
We believe in people and relationships with people.
Technology is part of that.
And we believe in educating our clients, helping them strategize.
The technology is there as something that helps them accelerate
something that empowers them and helps them take that strategy into their own hands.
At the end of the day, we are partners in technology
as much as we are partners in marketing strategy.
We're going to be a part of the process.
There will always be a human in the loop to ensure that when you're using the AI,
you're using it in a way that truly solves problems.
So your earlier point doesn't just create more problems for you
either because you floated your tech stack
or because you got your hopes up.
Or maybe even because there was bad data somewhere along the lines
and you kind of were consuming it in a vacuum.
Ooh, the vacuum.
I mean, the most dangerous thing in our industry
for a couple of reasons.
For those that forget to use it to literally vacuum up
the burnt popcorn kernels off the showroom floor.
But in this sense, can you tell I've been doing this for a while
and just couldn't give two boots?
OK, because it's the truth.
But also the vacuum you're talking about, Mac,
which is nothing worse than data analysis in a vacuum.
Oh, my gosh.
Very dangerous.
Very dangerous.
What are some of the things that you see have happened
as a result of data analysis in a vacuum?
It always kind of plays out.
There's there's different different circumstances.
But I see it play out the same way.
Is it over here?
We see a flashing red signal to our right.
It would be some sort of signal that's coming up in the deck, the data.
So we start looking over here.
And we are so fixated on it, blinders are on.
And so we are missing the holistic portrait that, you know,
that full picture of the data.
If I zoomed in really, really close right now,
I could just see your face, Michael.
Zoom out.
I see this whole showroom floor with all these wonderful people
and these exciting booths getting the full picture there.
Zooming in, you are missing out on so much.
And sometimes when you're super zoomed in,
you are seeing something that's flashing over here,
but you may be missing other indicators over here
that are actually very positive, that offset this negative to your right.
Or just something that tells a completely different story.
Right. So I think it's really important that we we zoom out.
We see the whole picture, the holistic portrait that the data is telling us.
And frankly, I think AI is really good at helping us see that whole picture
because it can ingest all the data and it can respond
and give it back to us very, very quickly.
And frankly, that's one of my favorite things about AI is to just get
all right, put all the cards on the table, give it to me straight.
I want to see everything good, bad and the ugly.
And oftentimes we see the good over here
and the bad or the ugly doesn't look so bad and ugly.
What is really staying out to me, too, is the ability then
to see it all in front of yourself, the human filter,
who can then factor in the nuance of the circumstances versus just kind
of taking the AI's word for it.
It's like I got the analysis piece done quickly,
but I need to ultimately be that filter set.
Absolutely. And hey, just remember, too,
that sometimes we we see those flashing red signals
and there is also an emotional reaction that happens.
We know this, it gets to be a little bit.
I know, I'm Italian, is unavoidable.
Yeah, you know, you know.
So I think it's really important to have that sort of piece of technology
in your arsenal that is emotionally unbiased as well.
Yeah, I love this conversation.
It's something to think about.
I love that, you know, there is the opportunity here to partner
with a company like Stream who has such a vast data set that becomes
the layer on top of the AI that actually drives
the diagnosis to help make better decisions.
As we close out, Mac, how can those listening or watching connect with you?
Connect with me on LinkedIn, the Kenzie Wiltrout.
You can find me VP of innovation at Stream Companies.
You can connect through the Stream Companies website
or you can send a carrier pigeon.
I am very open to meeting new friends.
How old? This is only send the carrier pigeon
when we're wanting to avoid Skynet, OK?
That's what we at least might be in the year 2026.
I don't know, but Kenzie Wiltrout.
Thanks so much for joining me on the Dealer Playbook podcast.
Michael, thank you so much for having me.
Hey, thanks for listening to the Dealer Playbook podcast.
If you enjoyed tuning in, please subscribe, share and hit that like button.
You can also join us and the BPD community on social media.
Check back next week for a new Dealer Playbook episode.
Thanks so much for joining.
About this episode
Mackenzie Wiltrout of Stream Companies breaks down what “impactful AI” should look like for dealerships amid the 2026 AI tool flood. She argues the differentiator isn’t the model—it’s human-in-the-loop implementation, dealer-specific focus, and proof via post-launch reporting/ROI. Generic, one-size-fits-all LLMs can reinforce bias and create “smoke and mirrors” if they don’t deliver measurable results. Stream’s approach centers on purpose-built AI in its Orange OS, paired with education and strategy support, plus a “zoom out” view to avoid data analysis in a vacuum.
The AI hype is real, and every vendor at NADA claims to have the next big thing for your car dealership. But as you consider integrating new tools, are you truly solving problems or just inviting new ones? The promise of AI in automotive retail is vast, yet without a strategic approach and "human in the loop," the technology can quickly become a costly distraction.
In this episode, you’ll discover:
Why generic AI solutions often fail to deliver real ROI in a dealership setting.
How to identify and choose AI tools that provide actionable data instead of just affirming biases.
The critical role of "organic intelligence" in leveraging AI to move more metal and improve fixed ops.
Strategies for avoiding common technological pitfalls that lead to change management nightmares.
Mackenzie Wiltrout, VP of Innovation at Stream Companies, shares her expert perspective on finding impactful AI that genuinely drives dealer growth.
Follow The Dealer Playbook so you never miss an episode.
