I sat down last week with Russell Sherwin. Former CMO of IBM Watson Commerce, former chief revenue officer at a PE-backed SaaS company, now a go-to-market advisor and a professor at the University of Georgia.

Twenty five minutes into a call that started as a scheduling mix-up, he said something that stuck.

"AI can't create. It can synthesize."

That's the whole tension right now.

Every founder and every competitor has the same tools, the same prompts, the same "best practices" AI can generate on command. So the question isn't who adopts fastest anymore.

It's who still thinks for themselves once everyone has the shortcut.

Russell told me about a former CMO at Gong who runs his brainstorms differently now.

He puts the question to AI first. Gets the top ten answers.

Then bans all ten from the meeting.

Why. Because AI is regression to the mean. It's every idea that's already been published, averaged into something safe. If your team starts there, they end there too.

Good marketers create. Great marketers steal, sure. But the value now sits entirely in what happens after you've cleared the obvious answers out of the room.

Most leaders write off culture as fluff. Russell used to be one of them, until he moved from managing a team to managing managers.

His framework for diagnosing an underperforming org, in order:

Are the goals right, given the market. Do we have the right strategy to hit them. Do we have the right people, organized the right way, executing that strategy.

Here's where most companies get stuck. They obsess over lagging indicators, revenue, closed deals, and never define the leading indicators that actually drive them: specific behaviors, specific activities. A rep can't control revenue. They can control how many of the right conversations they have this week.

If you haven't defined those behaviors, you can't tell if you have a people problem or a system problem. You're just guessing, and guessing usually ends in a bad hire decision that was never really about the hire.

The line that reframed my whole week: selling isn't telling your story. It's helping the buyer write theirs.

Most reps are trained to lead with features, history, differentiators. Russell's approach inverts it. Form a hypothesis about what a win looks like for them. Turn that hypothesis into questions. Let the questions do the educating.

If the win-win is real, the buyer arrives at your value on their own. Nobody had to be convinced of anything they didn't already believe.

Ask AI your team's next big question. Get its top answers. Rule every one of them out before the meeting even starts.

How to implement: Prompt the exact question your team is wrestling with. Print the top 10 answers. Open the meeting by ruling all 10 out loud.

Expected outcome: Your team stops anchoring on the obvious and starts arguing from first principles.

Pro tip: Do this monthly, not once. The value compounds as your team gets used to reaching past the AI baseline.

Common mistake: Skipping the AI step entirely and assuming your team is already original. Most teams are regressing to the mean without realizing it.

If you can't name the specific behavior that drives the outcome you want, you can't coach toward it, and you can't diagnose why it's missing.

How to implement: Pick one lagging indicator that matters, revenue, retention, whatever. Write down the 2 to 3 behaviors that, done consistently, would move it. Not outcomes. Behaviors. "5 discovery calls a week," not "more pipeline."

Expected outcome: You can now tell the difference between a coaching problem and a hiring problem.

Pro tip: Use skill versus will to sort it further. Missing skill means enable and coach. Missing will means it's a direct conversation, not a training plan.

Common mistake: Managing to the lagging indicator directly. It's the number everyone watches and the one nobody can actually influence day to day.

Draft the case for your offer entirely from the buyer's point of view, before you write a single line about your product.

How to implement: Write down what a win looks like for them specifically, in their language, not yours. Turn each point into a question you'd ask them. Lead the conversation with those questions instead of your deck.

Expected outcome: The buyer starts finishing your sentences instead of waiting for your pitch to end.

Pro tip: If you can't write their version convincingly, you don't understand them well enough to sell to them yet. That's useful information too.

Common mistake: Leading with relevance and differentiation but skipping defensibility. If a stranger has no reason to trust your claim yet, the first two don't matter.

"Good marketers create. Great marketers steal. AI just made stealing free. What happens after that is still up to you."

The leaders who separate from the pack this year won't be the ones with the best AI stack. They'll be the ones who know exactly which questions their AI can't answer for them, and who still do the work to answer those themselves.

What's one question in your business you've been letting AI answer that you should probably be answering yourself?

Kalei Poteat runs Fleet Connect and Scalematic, and interviews operators, engineers, and executives on Legacy Decoded about what actually separates performance from noise once the tools are the same for everyone.