It was a Saturday, open-to-close, and I was watching a vendor demo on my lunch break. The sales rep was showing me a dashboard. Beautiful thing. Charts, graphs, AI-generated insights, predictive analytics. He kept clicking through screens, pointing at numbers, telling me what the tool could "surface."
I asked one question: "Can it clear a VinSolutions task?"
He paused. "Well, it integrates with your CRM to pull data in."
"Right. But can it take action in VinSolutions? Can it close a task, send a text, update a record?"
"The insights would inform your team on what actions to take."
That's the whole problem with dealer AI tools right there. They show you what to do. They don't do it.
The Dashboard Problem
Vendors sell dashboards. Dealers need workflows.
A dashboard tells you that your internet response time is 34 minutes. Cool. I already knew it was bad. What I need is a system that alerts me when a lead has been sitting for 5 minutes so I can fix it in real time. The dashboard shows me the problem yesterday. The workflow fixes it today.
I've seen stores pay $1,500/month for AI dashboards that their managers open once a week. Maybe. The data sits there, looking pretty, while the same problems happen on the floor every single day. Nobody's workflow changed. Nobody's morning got easier. The tool just added one more login to ignore.
My morning report isn't a dashboard. It's a PDF that shows up on my phone at 6:15 AM with exactly what I need to know and exactly what's wrong. I don't log in to anything. I don't click through filters. The information comes to me, pre-analyzed, with the problems already flagged.
That's the difference between a dashboard and a workflow. One waits for you. The other works for you.
Why "Bolt-On" AI Always Fails
Most dealer AI tools are built by tech companies, not car people. They look at the dealership from the outside and see data problems. "If we could just aggregate all this data and apply machine learning..." Sure. But dealerships don't have data problems. They have process problems.
A store with a broken follow-up process doesn't need an AI to analyze the broken follow-up. It needs someone to fix the follow-up process. Then automate the fixed version.
I've watched this play out three times at stores I know:
The Predictive Lead Scoring Tool
One store bought a tool that scored internet leads from 1-100 based on "likelihood to purchase." The idea was that reps would prioritize high-scoring leads. In practice, reps ignored the scores completely. They called leads in the order they came in, same as always. The tool sat there scoring leads that nobody looked at.
The real problem wasn't lead prioritization. It was that the store had 3 BDC reps handling 200+ leads a month with no clear ownership rules. Leads fell through cracks because nobody knew whose job it was to call them, not because they couldn't tell which ones were good.
The AI Desking Tool
Another store tried an AI-powered desking tool that would recommend pencil numbers based on market data, inventory age, and customer credit profile. Sounds smart. The desk managers used it for about a week, then went back to DealerTrack and their own mental math.
Why? Because the tool didn't know that the Telluride on the lot for 47 days had a $1,200 dealer add-on that the customer was going to push back on. It didn't know the customer's trade was a 2019 Forte that wholesale was soft on this month. It didn't know the sales rep already quoted a payment over the phone that was $40 lower than what the tool recommended.
Context matters. AI tools that don't have the full picture make recommendations that experienced desk managers immediately override. So the tool becomes furniture.
The Automated Email Platform
Third store signed up for an AI email platform that would write "personalized" emails to every customer. The emails were technically personalized (they included the customer's name and the vehicle they looked at) but they all read like they were written by the same marketing intern. Because they were. The AI just swapped in variables.
After 90 days, the store's email open rate dropped from 12% to 6%. Customers learned to ignore the emails. Some marked them as spam. The store's sender reputation took a hit that affected their legitimate emails too.
Personalization isn't "Hi {first_name}, still thinking about the {vehicle}?" Personalization is knowing that this customer came in with her husband on Saturday, test drove the Sportage, loved the car but balked at the payment, and her husband mentioned they were also looking at a Tucson. A real follow-up references that conversation. An AI template can't.
What Actually Works Instead
Build on top of what you already have. Don't replace it.
Every dealership already runs a CRM. VinSolutions, DealerSocket, Elead, whatever. Your reps are in it every day. Your processes (good or bad) live there. Your customer data is there.
The smart move isn't buying a new platform that sits beside your CRM. It's automating actions inside your CRM. That means:
- Reading CRM data to make decisions (not exporting it to another tool)
- Taking actions in the CRM itself (clearing tasks, logging activities, triggering follow-up)
- Surfacing information where your team already looks (not in a separate app)
- Fitting into existing workflows instead of creating new ones nobody follows
That's what I built at Ancira. My automation logs into VinSolutions directly. It reads real CRM data. It takes real actions. My reps don't know there's an AI involved. They just see a clean task list, timely follow-ups, and a morning report that's already done when they walk in.
Fix the Process First
Before you spend a dollar on any AI tool, answer these questions:
- What is your follow-up process for unsold showroom visitors? Written down, not just "the reps know"
- What happens to an internet lead in the first 5 minutes? Who owns it?
- How do you decide which CRM tasks are real and which are noise?
- What does your morning meeting cover and where does the data come from?
- When a customer leaves without buying, who calls them back and when?
If you can't answer those clearly, no AI tool is going to help. You'll just be automating chaos.
I spent two months documenting every process at my store before I automated anything. The save-a-deal meeting format. The follow-up rules. The task triage logic. All on paper first. Then I built automation to execute those documented processes.
The automation is fast. Documenting the process took the real time. But that's the work that actually matters.
Vendors Sell Features. You Need Outcomes.
Next time a vendor shows you an AI tool, don't ask what it does. Ask what it finishes.
Does it finish follow-up, or does it start it and hand it back to your already-overwhelmed BDC? Does it finish the morning report, or does it give you a dashboard you have to interpret yourself? Does it finish task triage, or does it flag tasks for someone else to sort through?
The tools that work are the ones that complete the loop. Data in, decision made, action taken. No human in the middle translating insights into action. The whole point of automation is removing the middle step.
I'm not saying every dealer should build custom automation like I did. That takes time and a specific kind of stubbornness. But I am saying: don't buy a tool that gives you more work disguised as help. If it adds a login, adds a screen to check, adds a step to your morning, it's not automation. It's overhead.
If you want to see how I approach this differently (custom automation built on your existing CRM, not bolted on top), check out the AI Setup page. Or just book a call and we'll look at your setup together.