• This thread is just the tip of the iceberg.The people ahead of the curve aren't Googling for answers — they're already in here, having the conversations you haven't found yet. DealerRefresh is free.Get the full picture →

What have you replaced with AI?

This is illogical. 30 years of watching F&I should put you into a position where you are driving the conversation, not fishing for ideas.

IDEA: Start a new thread here on the forum. Pick your fav topic and start dropping wisdom bombs in it.
Joe, Not fishing for ideas, I am identifying issues that many need assistance with and solving them with that wisdom, using my experience of successfully training and guiding teams to successful outcomes, with all due respect. I jumped on this particular thread to lead, in that discussion and engage those that raise their hand in a space they should feel empowered to do so. No question is too small or too big in this business if it leads to profitable results.

I'll start a new thread, to drop some wisdom, knowledge and further discussion and find an audience, you are welcome to hop on anytime!
 
  • Like
Reactions: joe.pistell
Here's a short, casual reply:

I've replaced most of my first draft writing. Blog posts, email copy, social captions. AI gets me 80% there and I just polish the rest. Went from hours to minutes.

Also replaced a chunk of my frontend busywork. Generating repetitive components, writing basic CRUD, converting designs to tailwind. Not full replacement but easily 40-50% faster.

What I haven't replaced is architecture decisions and debugging complex issues. AI still falls apart when things get weird.

Curious if anyone has fully replaced customer support or internal documentation yet. Those seem close.
 
  • Like
Reactions: ErikJonker
Joe and Harrison both have good points here.

I've been building dealer-focused AI products for the past several months, and the pattern I keep seeing is that tools fail when they stack too many behavior changes at once: new login, new UI, DMS integration, staff retraining. The version that works is the one that lives inside something the dealer already uses, and produces output the dealer was already going to write.
You nailed it, Todd. The behavior change stack is exactly where dealer AI implementations fall apart.


On the fixed ops side, what we replaced was the mental load of the service advisor during check-in. Not the advisor. Not the write-up process. Just the part where they're supposed to simultaneously listen to the customer, remember the vehicle's service history, cross-reference open recalls, and mentally flag anything the last three visits might have missed. All before the customer walks to the waiting room.


That cognitive juggling act is where revenue leaks out quietly every single day. The advisors who are great at it are great because of experience, not process. The ones who are newer miss things. AI can close that gap without asking anyone to log into a new system or change how they write an RO.


I'm Carl Schiller, co-founder of MichigAIn Automotive Solutions. We're entering the pilot phase at a Michigan GM rooftop and the one thing we can say we genuinely replaced? The guesswork.
 
Pattern recognition, decisioning and service schedule optimization are the best use cases for AI.
Back in the day when roller hockey was a thing, I bought a pair of roller hockey skates. Think traditional hockey skates except with wheels. I considered myself pretty fast but everyone and I mean everyone was skating circles around me. They all had a good laugh then told me those are nice new skates. You will need to upgrade the bearings if you don't want to have a heart attack out here. We should be using the data and technology to optimize the human experience not replace it. Do you think there is a correlation between your google reviews and the fact that it takes 6-7 outbound messages to get someone to book service? What percentage of your chats are people who's car needs immediate service? Who monitors that? Did the LLM improve your show rate? close rate?
 

✨ AI Highlights

Dealer and vendor professionals share what AI has actually replaced in their workflows, with examples ranging from QA writing and UI/UX development to first-draft content and service advisor mental load during check-in. The thread's clearest insight is that AI works best when it fits inside existing processes rather than forcing behavior changes, and that getting good results requires active management — including knowing when and how to push back on the tool. A side tangent involving an F&I-focused poster and a nostalgic archive screenshot of Alex Snyder's 2005 dealer website add texture but modest signal.

Replies Views 25 7,295 Started Last Reply