President and CEO of CarEdge, Zach Shefska opened the first keynote at the new AutoIndustry.ai Summit by launching 100 AI agents to mystery-shop 100 dealerships live on stage. Within the first hour, 83 stores had responded and 12 had already sent out-the-door quotes.
It was a live look at how the shopping process is already changing.
Zach also shared the broader scale behind it. CarEdge has deployed more than 100,000 AI agents since July 2025, generating over 1.5M messages between consumer-side AI agents and dealership teams, touching more than $500M in negotiated vehicle sales and saving customers $22.7M.
Those numbers are worth paying attention to.
AI Exposes Pricing Gaps at Scale
In one CarEdge experiment, 100 Ford dealers quoted nearly identical F-150s. The spread between the lowest and highest out-the-door quote reached $9,256 before sales tax.
An AI agent can surface that spread in minutes.
A store can no longer assume the customer will only see one version of the market. The customer now has access to software that compares dozens of quotes, organizes them, flags the add-ons, tracks doc fees, and notices when the online price and the real quote drift apart.
Zach said his team captures more than 160 data points in every interaction. That means the store’s pricing behavior is no longer just a sales issue. It’s now a data issue, a reputation issue, and a conversion issue all at once.
Clean Operators Gain More Than Goodwill
One of the most useful parts of Zach’s keynote came from consumer reaction. When CarEdge published mystery-shop results, shoppers asked to see which dealers quoted honestly so they could buy from them.
That matters more when the data shows most dealers are already doing solid work. Across nearly 60,000 out-the-door quotes, 60% of dealers earned an A for transparency. Only 3% earned an F.
The opportunity is right there for stores that quote cleanly and stay aligned from listing to deal.
Zach Shefska
A New Retail Channel
The more important idea in Zach’s keynote had nothing to do with a cool live demo. It had to do with what this creates inside the dealership.
AI shopping agents create a new channel. Traditional walk-in traffic still exists. Regular internet leads still exist. Now a third kind of demand is forming, where the first layer of the transaction gets handled by software. Search, comparison, first quote, negotiation logic, and early handoff all start changing shape. Zach described that as AI-enabled retail, and he already sees stores beginning to build around it.
That matters because a dealership now has to decide what part of the process it wants humans to own and what part it wants systems to carry.
A store with clean inventory data, defined pricing rules, good CRM hygiene, and a disciplined quote process will handle AI traffic very differently than a store where every quote depends on who opened the lead and how busy the desk happened to be.
That’s where most stores will either build trust or burn it.
Dealer Playbook
1. Secret shop your own store with AI.
Run the same test Zach ran. Compare the advertised price, first response, out-the-door quote, and final pencil. Look for drift, delay, and surprise. Then fix it.
2. Audit how your team answers pricing questions.
Pull recent leads and review how often customers asked for a number, how often they got a direct answer, and how often the answer actually matched what they saw online.
3. Clean up one pricing friction point this quarter.
Pick one: Add-ons, doc fee language, quote formatting, listing-to-quote alignment. Remove one point of confusion customers keep tripping over.
4. Fix the data before adding more tools.
Inventory fields, CRM notes, pricing tables, and lead-source tagging all need to be accurate. AI will amplify whatever sits underneath.
5. Put one manager in charge of AI readiness.
Someone should test the website, lead flow, response process, and quote quality the way an AI shopper would. Ownership speeds everything up.
