For thirty years as a founder and CEO, I've been obsessed with a simple question: What's on the back of my shirt?
Not literally, of course. It's a metaphor. The stuff everyone around me can see that I can't. The blind spots. The habits. The tells. The patterns that are obvious to everyone except the person who's supposed to be leading the team.
I've tried everything to create psychological safety so people would tell me the truth. Open door policies. Anonymous feedback forums. "Challenge me anytime, I can take it." I've explicitly invited criticism. I've told people I'd rather hear hard truth than comfortable lies.
Here's what I learned: that doesn't work.
People don't want to criticize the person who signs their checks. No matter how safe you make the environment. It's human nature, and I don't blame them for it. Your VP of Sales isn't going to tell you that your discovery calls are weak because she needs you to be confident in the partnership. Your engineer isn't going to say "you don't understand our architecture" because you're the guy deciding whether to fund his next feature. Your CFO isn't going to say "that business decision was a mistake" because he needs your trust to do his job.
The incentive structure is just too strong. Safety isn't enough to overcome it.
The unlock for me has been AI.
I now run the transcript of every important sales call, every partnership conversation, every board interaction through Claude and ask for brutal coaching feedback. McKinsey-level critique. Don't hold back. Grade my performance. Tell me where I went wrong.
And yes - you have to prompt past the AI's own default friendliness. If you just ask "how did I do?" the AI will tell you everything was great. It's being helpful. But if you prompt correctly, asking for specific, harsh feedback with evidence, the insights are devastating. In the best way possible.
Last week I got coaching feedback on a ninety-minute partnership meeting with a potential customer. The grade: B minus. The feedback: "You gave away your entire roadmap to them without getting any commercial commitment. The meeting generated excitement but no signed agreement. You're acting like you need this deal more than they do. A stronger closer would have qualified their budget and timeline before sharing roadmap details."
Ouch. Completely accurate. And something no human on my team would have said.
Not because they don't see it - they absolutely do. But because they're human beings with job security that depends on me trusting them. Even the strongest team leads have that bias.
Here's what makes AI different: the AI has zero incentive to be nice. It doesn't worry about its next performance review. It doesn't need a bonus. It doesn't benefit from making you feel good. It benefits from being useful - and being useful means telling you what you actually need to hear.
The prompt provides all the psychological safety the AI needs to do that. It doesn't worry about your feelings. It just gives you the truth.
This matters more as you scale. When you're a founder, you get feedback from the market - customers tell you what's working and what's not. When you're a CEO managing a team, you lose that direct feedback loop. Your team gets filtered through incentives. Your board tells you what they think you need to hear. Your investors are friendly because they want to believe in you.
But your sales calls? Your partnership conversations? Your board presentations? Those have witnesses who know what you're actually doing. Those conversations have transcripts now. And if you're willing to ask an AI to tell you what's actually going wrong, you can see what everyone else sees.
The back of your shirt is still there. Most people just won't tell you what it says.
But an AI will.
