Every company runs two deals at once.
There's the one you say out loud. We value your thinking. We want you to grow here. We're building something together. Then there's the one you actually run. You're paid for output. You're measured on volume. You're replaceable by whoever produces more of it.
Everybody signed the first one and lived the second one. And honestly, that was fine. Output was expensive, so nobody had to look at the gap.
Then your team started using AI. Not the way you rolled it out. Their way. Quietly, at home, at 11pm, on the parts of the job they hate. And inside a month they found out something you'd have preferred they didn't: the thing you pay them for can now be produced in minutes.
The whole talk, in one line
AI isn't breaking your employment contract. It's reading it out loud.
What your best person worked out before you did
If they hand you the leverage, you take it. The work gets faster, the headcount conversation gets easier, and their pay stays exactly where it was. They ran that math in about four seconds. It isn't disloyalty. It's arithmetic.
So they use it and they don't mention it. Your most valuable employee is now running a private productivity gain you can't see, can't manage, and can't buy. And the first company that offers to pay for judgment instead of output gets them, because they already know what they're worth. They asked.
Why the rollout didn't move the needle
You gave everyone a license. You ran the training. Adoption looked fine on the dashboard. Nothing changed in the numbers, and nobody could tell you why.
Here's why. Nobody is going to 10x for a salary. Not because your people are lazy, but because you never asked them to and you certainly never offered to pay for it. The rollout was a tools problem. The gap is a contract problem. You can buy every seat on the market and it won't touch the second one.
The three renegotiations
These aren’t repairs. Nothing is broken that wasn’t always this way. This is you writing the real contract on purpose, out loud, for the first time, while it still counts as leadership instead of damage control.
- 1What you permitRight now the safe play for your best people is to hide the leverage, because the upside is theirs and the downside is yours. You need an amnesty, and it has to be worth something. Not a policy memo. A price. Until showing you what they built pays better than hiding it, they'll keep hiding it, and they'll be right to.
- 2What you pay forYou're buying deliverables in a market where deliverables cost close to nothing. Start buying the things that didn't get cheap: judgment, taste, the willingness to own an outcome nobody scoped. Then say it out loud in a comp conversation. If you don't say it, nobody believes it, and the stated deal stays fiction.
- 3What you countHours and volume were a fair scoreboard when volume was hard to produce. It's free now, so it measures nothing. The number that survives is leverage: what did this person build that keeps working when they aren't in the room. Count that, and the whole company reorganizes around it without you sending a memo.
What the room walks out with
- The two-deal read on your own company, in your own words, written during the session.
- The specific reason your AI rollout produced adoption without productivity, which is almost never the reason you were told.
- An amnesty you can actually run on Monday, including what it costs and what you get back.
- The one sentence to say in your next comp conversation that changes what you're buying.
- A short, uncomfortable list of who in your company is already 10x and hasn't told you.
Two ways to run it
The keynote
20 to 25 minutes, plus Q&A
Built for CEO groups, life-science and founder seminars, expert webinars, Vistage rooms, and conference stages. One argument, delivered straight, with room left for the fight afterward. Virtual or in person.
The workshop
45 to 90 minutes, one leadership team
Same argument, but we stop diagnosing and start writing. Your team drafts its own three renegotiations in the room, out loud, in front of each other. That's the part that doesn't survive being emailed around later, which is exactly why it works.
Who should be in the room
- CEOs and founders at companies between $5M and $75M, where one person can still change the deal without a committee.
- Leadership teams who bought AI seats last year and can't point at what changed.
- Technical founders who'd rather solve this with tooling and are about to find out it isn't a tooling problem.
- Rooms that can handle being told something true before they're told something comfortable.
Who’s giving it
Greg Rosner, messaging therapist for B2B founders.
He wrote StoryCraft for Disruptors, created the Magnetic Messaging Framework, and runs Claude Training for CEOs, a working group where founders build with AI in front of each other every week.
He isn’t a technical guy. He runs eight AI agents that scout signals, write, run outreach, watch his systems, and hand him a morning brief. He built all of it himself, with no engineering background. That’s not a credential he’s bragging about. It’s the reason the talk lands: he is the employee who went 10x, and he knows exactly what it would have taken to get him to tell his boss.
This is just truth. The only question is whether your team hears it from you, or from the company that hires them next.