How Auto Dealers Are Building Custom AI Tools In-House, Starting With What Annoys Them Most

Car dealerships are no longer just buying off-the-shelf AI. A growing number are building their own tools, and three practitioners share the beliefs that are making it work.

Belief 1: Complaints Are a Feature, Not a Problem

Chris Hudson, general manager at Mark Miller Subaru in Salt Lake City, launched a program he calls the "Frustration Harvest." He ran a $100 gift-card contest asking staff to submit their biggest workplace complaints. Those complaints became the roadmap for in-house AI projects. The Frustration Harvest now lives inside Jarvis, the dealership's broader operating system. "Pick one thing you're paying a vendor for that bugs you, and build it," Hudson said. Kevin Pitts of Tom Masano Auto Group in Pennsylvania uses the same logic, building a marketing accountability tool that connects to Google and social data and generates a Monday-morning report, a recon photo tool that automates inventory image publishing, a financial reporting tool fed by general-ledger data, and a driver dispatch replacement tool currently in progress.

Belief 2: You Learn More by Starting Than by Waiting

Mike Yates, general manager at BMW of Bridgewater in New Jersey, recommends pasting a real document into an AI tool and seeing what it catches. When he tried it on an office-space lease, the AI found an unfavorable termination clause and overpriced maintenance fees. His advice: use a paid account (not a free one), train the AI to understand your context before asking it for anything, and connect your own data before trusting outside reports.

Belief 3: Treat the First Build as Version One, Not the Final Version

None of the practitioners treat early tools as finished. Pitts reviews every AI-generated report before it goes anywhere. Hudson encourages staff to demo their own builds and formalize the viable ones into the broader system.

Key takeaways:

• Starting with employee pain points produces immediately useful tools and faster internal adoption
• Direct data integration (GA4, general ledger, inventory feeds) drives measurable accountability outputs
• Human review checkpoints remain standard practice, especially for customer-facing AI outputs

Read the full article on Car Dealership Guy News