Over twenty pieces we’ve talked about how to make AI work in your business. How to onboard it. How to measure it. How to fix it when it breaks. How to build agents that don’t hallucinate. How to rethink your whole org chart around it.
That’s all real. All necessary.
But there’s a recursive insight underneath all of it that I didn’t think to name until we got to the end. And now I can’t unsee it.
Here’s the thing — the reason you’re struggling with AI is the same reason you’re struggling with your team.
Not “AI is like your team.” That’s soft and circular. I mean — you have a habit. A specific one. And it’s costing you twice. Once with people, once with machines. But you see them as two separate problems. They’re not. They’re literally the same problem wearing different clothes.
Two delegation stories
I worked with an HVAC owner last month. Two direct reports. And he described the same pattern ten other owners have brought to me in the last six weeks.
First report. Owner gave detailed instructions. Step one, step two, step three. Email the customer at this point. Use this template. Write back like this. Micro-managed. Report is competent. Capable. But the owner had removed every decision from the work. “Follow these steps” was the whole job. Report felt insulted. Went quiet. Unmotivated. Owner concluded — “I hired someone who doesn’t have initiative.”
Second report. Same owner, different approach. Owner said — “I need all our new customers to feel like we’ve solved their problem even before the first appointment. How do we make that happen?” No steps. Just outcome. Then got out of the way. Report called me six weeks in — “I’ve never had more fun at work.” Report invented a three-point follow-up sequence, rewrote the intake form, changed when they ask about budget. Owner barely micromanaged at all. Report took ownership.
Same owner. Same skill. Different approach. One hired judgment then withheld it. One hired judgment and used it.
Now the same owner tries the AI on a customer follow-up email. Writes a prompt — “Write an email to customer confirming their Tuesday appointment. Tell them we’ll arrive between 9 and 11. Apologize for the delay in getting back to them. Ask if they have any questions. Sign it with my name. Use a friendly tone.”
Runs it. Reads the output. Thinks — “This is generic. Doesn’t sound like us. AI isn’t ready.”
Deletes it. Writes the email himself.
I ask — “Did you tell it what the outcome should feel like?”
Owner — “I gave it all the steps.”
“Did you tell it what ‘good’ looks like?”
Silence.
“Have you shown it an email you’d be proud of?”
More silence.
Yeah. He had given the AI the same thing he gave report number one. Detailed steps. Zero judgment. Then concluded the tool wasn’t smart enough.
The pattern
Here’s what I saw. Over-specify the steps. Under-specify the outcome. Hire judgment. Then withhold the judgment.
Owner does this with report one — “here’s exactly how.” Owner does it with the AI — “here are the steps. Make it work.” Same move. Different context. Same mistake.
Pattern is recursive. It repeats.
When you delegate to a person and over-specify steps, you get compliance. You don’t get thinking. You get someone following a recipe.
When you delegate to AI and over-specify steps, you get the exact same thing. Compliance. Output that follows the recipe. Nothing that smells like judgment.
But the diagnosis feels different. With the person, the owner thinks “they don’t have initiative.” With the AI, the owner thinks “the tool isn’t sophisticated enough.”
Both are wrong. Both are the same mistake.
Why it matters
This is costing you twice over.
With your team, over-specification kills motivation. You get people who do what they’re told and nothing more. You have to manage everything. You can’t scale because every decision goes through you. The pyramid model we talked about in piece #19 — it exists because owners over-specify and under-delegate. Structure becomes a proxy for the decision-making you wouldn’t trust to anyone else.
With AI, over-specification kills output quality. You get generic email templates. Mediocre blog drafts. AI output that sounds like every other AI output. You get “it’s good for basic stuff but nothing important.” Which sounds like a tool problem. It’s a delegation problem. You told it how, not why. You never showed it what good looks like.
Cascade is the same in both cases. Owner decides what “good” is and keeps it private. Owner specifies every step to get there. Owner concludes the agent (person or machine) doesn’t have the judgment to do it right. Owner keeps doing it themselves or manages tighter.
Business gets stuck. Owner stays ON the business, not IN it.
The recursive insight
Here’s the thing that changes everything.
Framework that fixes your delegation to people is literally the framework that fixes your delegation to AI.
Ingredient Five of the Professional Recipe is called Output Over Process. It says — describe what good looks like, not the steps to get there. Let the agent (human or machine) figure out the path.
That’s not an AI principle. It’s a management principle. Works with people because people are smart enough to interpret a direction and find the steps. Works with AI for exactly the same reason — when you describe the outcome, the intelligence (human or artificial) can solve for the path.
You’ve probably been delegating to people your whole career. You’ve probably been doing it wrong in exactly this way — over-specifying steps, under-specifying outcome. You’re used to it. It feels safe. You know what you’re going to get.
Now you’re delegating to AI. And the exact same habit that’s been breaking your team is now breaking your AI deployment.
Recursive insight is this — fix the habit once, and you’ve fixed it everywhere.
If you learn to delegate to your team by describing outcomes instead of steps, you suddenly become better at delegating to AI. If you learn to onboard AI by being crystal clear about what good looks like, you suddenly start managing your team differently. Skill transfers. It’s the same skill. Context changes. Principle doesn’t.
Does that make sense?
The Monday Move
Pick a delegation you made to someone on your team in the last week.
Look at how you described the work. Did you tell them the steps? Or did you describe what good looks like?
Now go back to an AI task you’ve prompted in the last week. Customer follow-up email. Sales pipeline summary. Competitor analysis. How did you describe that task?
Do you see the same pattern?
If yes — if you over-specified steps in both cases — you’ve found your leverage point. Fix this one habit and you’ve fixed a bottleneck that’s been invisible to you.
This week — pick one AI task you’re about to do. Instead of writing “steps,” write “outcome.” Tell it what good looks like. Show it an example if you have one. Then run it.
Do the same with one delegation to a person.
Watch what changes.
The series closer
Twenty pieces ago, we started with the idea that you hire smart people and then don’t trust them. You specify every step because you’re afraid of the wrong output. You end up managing tighter and scaling slower.
Then we talked about AI the same way — like a tool you have to micromanage. We talked about how to build agents that work. How to measure them. How to fix them. How to rethink your whole business structure around them.
All of that was real.
But the insight that ties it all together — the one that makes the framework portable to everything you delegate — is this. Framework is the same because the principle is the same.
You’re not bad at AI. You’re bad at delegating.
And bad at delegating isn’t permanent. It’s a habit. You can see it. You can change it. And once you do, you’ve fixed it everywhere.
Skill is the same. Context changes. Get good at delegating to one, you’ve gotten good at delegating to both.
That’s what the twenty pieces pointed to. Not tactics. Not frameworks bolted onto your process. A recursive pattern about how you work. Pattern you’ve been living with for years, now visible. Now changeable.
So. Fix it once. Fix it everywhere.
Framework: The Professional Recipe — Ingredient 5: Output Over Process (the core discipline). See also: Stop Hiring AI. Start Building It. (the verb shift from HR to Engineering). From Pyramid to Station Plan (the structural pattern this recursive habit creates).
