The Most Important KPI of the AI Era

Aug 10

This weekend I was talking with a world-class engineer. True story.

A few months ago, his firm laid him off. He'd done everything right: he shipped, he mentored, he delivered. Didn't matter. Then, of course, the firm discovered what every firm is discovering right now: AI can't manage AI. Someone still has to frame the work, inspect the output, and know when things are going sideways. So they brought him back as a contractor, and today he spends his days orchestrating a fleet of AI agents doing engineering work.

I asked how it was going.

"Well, I'm making good money," he said. "But honestly, I'm concerned. I'd love a real job with a great team. I've just... never been good at sales. Especially sales of myself! So it’s been hard to find a permanent role."

I asked him one question:

"How have you approached the problem of selling yourself — an AI-augmented engineer — with AI augmentation?"

He blinked. "Wait. What do you mean?"

"You manage AI agents to perform engineering tasks, right?"

"Yes."

"Are you managing AI agents to perform sell-yourself tasks?"

Long pause. And you could almost watch little parts of his brain growing in real time (like tiny mushrooms after rain).

"I've... never thought about that."

That's it. That's the whole thing. Here's a man who spends eight hours a day directing artificial intelligence, and yet it had never occurred to him to point that same capability at the problem keeping him up at night.

The technology wasn't the limitation. The thought was.

The KPI nobody's tracking

When I speak to leadership teams, I like to tell them the most important KPI of the AI era is TTFS.

Then I ask: "Anybody know what TTFS stands for?"

Nobody does. (Because I made it up.)

TTFS: Time To Forehead Slap.

TTFS measures the distance between knowing AI can help you and remembering to let it. It's the time that elapses between the moment AI could have joined the work and the moment you smack your forehead and go, "I can't believe I didn't think to involve AI in this."

I've written before about wherewithal moments — those flashes where you suddenly realize, mid-struggle, that you've been doing something alone that you didn't need to do alone. A wherewithal moment is the event.

TTFS is how you measure it.

Importantly, the KPI isn't "number of times you forgot AI existed." That number is high for everyone. The KPI is how long it takes you to remember.

We are all slapping our foreheads

You might think this only happens to novices. I wish.

Arnfinn is one of the most forward-leaning members of my AI Junto — the kind of person who experiments more in a week than most professionals do in a quarter. Recently he posed a thorny question to the group, and someone replied, a little sheepishly, "Have you asked AI?"

Arnfinn's entire response was the forehead-slap emoji. 🤦

One of the most advanced practitioners in the room hadn't thought to do the most basic thing.

Now for my own confession. This one's embarrassing.

I'm currently onboarding an executive leadership team that's working to become AI-native, doing one-on-one calls with every member of the ELT. This week, the CFO mentioned that he and his wife do their family budgeting with AI (shocker!!).

Forehead. Slap. (My own.)

Because just last night, my wife and I had sat down to talk about our family finances. And it did not even dawn on me to invite AI into the conversation.

Let that sink in. I make my living teaching people how to work with AI. And in a conversation with my wife about how to be good stewards of the money we're making from AI... I (the "AI expert") never thought of bringing AI to the table. Not till this morning, when the CFO shares that as the example in his ELT onboarding interview.

The register I hope you hear in all of this: embarrassment without shame. The little private humiliation every competent person is experiencing right now (that the help was available, the team was standing by, and we still defaulted to artisanal struggle) isn't evidence you're bad at AI.

It's evidence you're mid-rewire.

If it happens to Arnfinn, and it happens to me, it's going to happen to you. The limitation is not the technology. The limitation is our imagination. We have spent our entire lives conditioned to what I've called unassisted living — handmade, artisanal cognitive work, done alone. The default is deep. It doesn't disappear because you got a ChatGPT license. It doesn't even disappear because you've spent years teaching it.

Measure it. Don't judge it.

Here's why I insist on framing this as a KPI and not a character flaw.

A few weeks ago I asked, What have you never measured? — because you can't see what you've never named. Well, this is me taking my own medicine. I wrote a whole book arguing that ideaflow is the only business metric that matters for innovation. TTFS is its sequel for the AI era: the lag I most needed to name in my own life.

And here's what naming it as a metric buys you that judgment never will: a metric implies a trend line. You don't judge a number. You move it.

Judgment, on the other hand, freezes learning. The moment "I forgot to use AI" becomes evidence that I'm behind, I'm failing, I'm a fraud, curiosity collapses. And curiosity is the whole game.

So here's the reframe: going from AI novice to AI native does not mean you stop forgetting. It means the interval shrinks.

Bad TTFS is three days after the deadline: "I really should have used AI for that."

Better TTFS is thirty minutes in: "Wait. Why am I doing this alone?"

Best TTFS is before you start: "What part of this should not be handmade?"

Nobody gets to zero forever. The pros just recover faster… and recover earlier each time. As I tell audiences: as long as your TTFS is decreasing, you’re probably on the right path.

A useful trigger while you're recalibrating — the post-it version of this whole essay:

Any time you wish you had 30 extra IQ points on a problem, that's your cue.

A negotiation, a budget, a job search, a hard conversation with a colleague. That wish is the forehead slap, arriving early.

Other people shorten the lag

Here's the hard part: you cannot slap your own forehead about something you never thought of. That's the whole problem. As Thomas Schelling put it, no matter how heroic your imagination, you can never think of that which would not occur to you.

Which is why the single fastest way to shrink your TTFS is other people.

When the CFO mentioned family budgeting, his realization became mine — his forehead slap saved me weeks of not-thinking-of-it. When someone sheepishly asked Arnfinn, "Have you asked AI?", that question was a gift, and Arnfinn's 🤦 was a gift right back to everyone watching: permission to admit we are all still adjusting. That's precisely what a community of practice does — every shared slap is a shortcut somebody else doesn't have to discover the hard way. (It's the reason the AI Junto exists, and why members share their experiments — and their misses — every week.)

We need to normalize the forehead slap. Not as a confession of inadequacy, but as a data point. The teams that get AI-native fastest won't be the ones that never forget. They'll be the ones that celebrate remembering — out loud, together.

This week's experiment

Keep a TTFS log for one week. It takes ten seconds per entry:

Every time you catch yourself thinking "I can't believe I didn't think to use AI for this," write down both the task, and roughly how long you'd been grinding on it unassisted before the slap arrived.
At the end of the week, look at your intervals. Don't judge them. Just notice them. That's your baseline.
Then share your best forehead slap with somebody else. Could be your team, your spouse, your group chat. Watch what happens: your slap becomes their shortcut, and theirs becomes yours.

The goal isn't to stop slapping your forehead.

The goal is to slap it sooner.

Join over 54,147 creators and leaders who read Methods of the Masters each week.

AIPersonal StoriesCommunityJuntoGrowth

Jeremy Utley