The GM Playbook: Leading AI Adoption Without Becoming the IT Department
QoreAI Editorial
Updated June 2026 · 3 min read
If the GM does not lead AI, it becomes a side project run by whoever has time, which is no one. This guide is for GMs who want to lead without becoming the dealership IT department.
What the GM owns
The decision of which three plays run this cycle. Not the dealer principal, not the marketing vendor. The GM, because the GM is the one who can pull the trigger on staffing and process changes.
The weekly cadence. Fifteen minutes a week on AI in the existing manager meeting. Not a separate AI committee. Anything that becomes its own meeting dies.
The kill decisions. Owning the call to kill a play that is not working is how you get permission from the team to try the next one.
What the GM delegates
Tool selection inside a play. If the marketing manager wants to run the service follow up play on Claude rather than ChatGPT, that is their call.
Day to day execution. The owner of the play runs it. The GM checks the milestone.
Training. Pick a champion in each department and let them train their peers.
What the GM ignores
Vendor pitches that do not map to one of the three current plays. The answer is no for ninety days.
Industry hype. The latest model release does not change your plan for this quarter.
Cross dealership consulting offers that promise to do it for you. The point of the 90 day cycle is that your team learns by doing.
Ready to benchmark your AI skills?
Get a personalized readiness score and see where your store actually stands.
FAQ
What if my dealer principal disagrees with the play selection?
Loop them in before lock, not after. Show them the assessment score and the rationale for each play. Disagreement at lock is fine. Disagreement at day 60 is fatal.
How do I handle a manager who refuses to use AI tools?
Make their KPI the lever, not the tool. If they hit the KPI without AI, fine. If they miss it, the conversation is about the KPI.
Should I send my team to AI training?
Only after they have run a play. Training without execution context does not stick.