BLiNK AI's Dave Perry has spent his career solving mission-critical problems, from secure communications in the Air Force to leadership roles across Fortune 500 tech. His read on automotive AI is blunt: most dealers are buying a tool before they've actually defined the problem it's supposed to solve.
Perry didn't come up through automotive. He came up through the Air Force, then Texas Instruments, then a string of tech leadership roles, before landing in the industry almost by accident. What surprised him most wasn't the technology. It was the structure.
A lot of other industries are more like democratic... but in the automotive industry, it's more like kingdoms.
OEMs hold power. DMS providers hold power. State franchise law holds power. Dealer groups hold their own power on top of that. Nothing in this industry gets solved by having the single best idea, because no one player controls the whole board.
🎙️ Want the full conversation?
Listen to this Auto Collabs episode for the complete discussion, additional context, and the conversations that shaped our perspective.
That structure is exactly why Perry thinks most AI purchases in this industry fail before they start. Here are five tests worth running before your next one.
Can You Explain the Problem in One Sentence, With a Dollar Figure Attached?
Perry's filter is blunt: if you can't quantify a problem in dollars and cents, it probably isn't your real priority, no matter how urgent it feels day to day.
If you can't describe it in dollars and cents, it may not be really worth solving.
Could a Seven-Year-Old Understand What You're Trying to Fix?
Perry references a program called Pitch a Kid, where founders have to explain their business to a room of actual second graders. If the explanation survives that, it's clear enough to act on. If it doesn't, the problem probably isn't defined yet, it's just described in jargon that's hiding the gap.
Are You Solving One Piece, or the Whole Value Chain?
Because power in this industry is scattered across OEMs, DMS providers, regulators, and dealer groups, a fix that only touches one piece of that chain tends to quietly fail everywhere else. Perry's framing: don't just solve a problem, solve it in a way that works for the OEM, the dealer group, and the DMS relationship all at once.
Is This a System of Record, or a System of Action?
This is Perry's sharpest technical distinction, and arguably the most useful one here. Your DMS, CRM, and CDP are systems of record. They store data. They don't act on it.
❝ It's not a toaster. It's not something you buy, plug in the wall, and now you've got it. ❝
Whatever tool you're evaluating, ask directly whether it's actually taking the next action on your behalf, or just giving you another dashboard showing you a problem you already knew about.
Does This Scale Your People, or Replace Them?
Perry's clearest example here is BDC pay structure. Most reps get compensated on activity, calls made, not on outcomes. His pitch: pay for the outcome, and let the rep use automation to get there faster.
Me working alone, I might book 15 calls a day. Me with the right tool, I could book 50.
That's the difference between AI as a threat to a job and AI as a multiplier on a person already doing that job well.
Why This Matters Most in Fixed Ops Right Now
Perry's case for urgency centers on one number: fixed ops represents roughly 15% of a dealership's revenue, but about half its profit. New car margin has been squeezed flat by pricing transparency. Service hasn't.
He also points to something dealers rarely track deliberately: a service advisor gets roughly ten times the customer touchpoints a salesperson does, and is statistically more likely to influence the next vehicle purchase than the person who sold the car in the first place.
❝ Every time I ask you another question or give you another option, your chances of booking that appointment go down by 7%. ❝
That single stat is a test all on its own. Every extra click, upsell prompt, or optional add-on in your own scheduling flow is quietly costing you completed bookings, and the customers behind them.
What This Means for Your Store
Run your next AI vendor conversation through these five tests using a real, current fixed-ops problem, not a hypothetical. If the tool fails even one, either the tool isn't ready, or the problem underneath it isn't actually defined yet.
Want help keeping up with all of this?
We run a weekly email dedicated entirely to AI in automotive, tracking what's actually worth your attention between the big splashy announcements. Sign up at AutoIndustry.AI.
