There is a story that former Tesla President Jon McNeill has been telling in recent months that has produced one of the most useful executive frameworks for the AI transformation moment.
The setup is 2015. Tesla was trying to sell $100,000 cars primarily online. McNeill was hired as President specifically to make the online sales operation work.
When McNeill dug into the actual online purchase experience, he discovered that buying a Tesla online at that time required 64 clicks to complete.
The Domino’s benchmark
McNeill and Musk were brainstorming this problem one day. Musk asked McNeill to pull up the Domino’s app.
“How many taps until you get a pizza?”
The answer was 10.
Musk’s response: “Let’s try to do this in ten.”
The 44 clicks nobody had questioned
McNeill’s team went and broke down the 64 clicks systematically. The pattern was clear. 44 of the 64 clicks came from loan and lease documents.
Every American who has ever bought a car through traditional financing knows what this means. Pages of loan documents. Disclosures. Consent forms. Truth in Lending Act notices. Privacy disclosures. And on. Every one existed for a defensible reason. Every one had been added by a lawyer trying to protect a bank from a specific risk.
McNeill did what almost nobody had done. He asked the actual question.
Are these legal requirements, regulatory requirements — or not?
The Talmud
McNeill’s description of what he got back is the sharpest line in this entire story. He described the loan documents as “like the Talmud” — accumulated over time by well-meaning lawyers, each one trying to minimize risk for their bank, each addition individually defensible.
“Most of this stuff, almost none of it is a legal requirement.”
The four-sentence reduction
McNeill’s reasoning: “If I’m buying a car, I’m essentially agreeing to a price. I’m agreeing to an interest rate and a payment for a period of time. So there’s four things. We ought to be able to describe that in a four-sentence paragraph.”
Four elements. Price. Interest rate. Payment. Duration. That’s what a car loan actually is at its irreducible core.
The Minneapolis moment
They went to the banks with a proposal for a one-paragraph loan document. Most refused. Then they got to one Midwest bank CEO — US Bank in Minneapolis.
He said, essentially: “I’m willing to give this a try.”
That single decision — by one bank CEO, in one Midwestern city — is why buying a Model 3 today takes about five clicks instead of 64.
The AI parallel
Every enterprise AI initiative I have visibility into is drowning in Talmud.
Governance frameworks. Approval chains. AI vendor review processes. Model validation requirements. Data classification workflows. Bias audits. Prompt-injection assessments. AI ethics committee sign-offs. Legal reviews of every use case.
Every one added by a well-meaning person trying to minimize risk. Every one individually defensible. Almost none of them actual legal requirements.
Nobody asks McNeill’s question. Is this actually a legal requirement — or just something we’ve built up?
Three practical questions
One: for each governance requirement in your AI deployment process, can someone specifically name the legal or regulatory basis for it? Not the risk. The specific statute, regulation, or contractual obligation. In most cases, no such basis exists.
Two: what is the AI use case actually — at its irreducible core? McNeill reduced a car loan to four elements. What is your AI use case at its actual core? Usually simpler than the current process implies.
Three: who is your Minneapolis bank CEO? Someone in your organization has the authority to say “I’m willing to give this a try” on radical simplification. Find that person.
The closing thought
McNeill’s insight was that the pile was not law. It was accumulated convention treated as law.
The question he asked — is this actually required, or is this just the Talmud? — is the single most useful executive question available in 2026.
Almost nobody is asking it about AI.
The world has changed. The leaders who notice will be the ones the next decade is built around.
