Save your seat →

Almost everyone I talk to is worried AI is coming for their job.

Almost nobody is doing the thing that would make them harder to replace.

Not learning another tool. Reps. In rooms. In front of people who can react in real time, where the whole value is that you're standing there and a model isn't.

That's the bet I've made, and speaking is the version of it I've chosen. I'm going to tell you what eleven days of that did, then show you the same move applied to your actual job.

---

Seven talks in eleven days.

The talk is called Hard or Wrong. It teaches one decision: whether the thing in front of you is hard, which means keep going, or wrong, which means change it. From the inside those feel identical, and picking the wrong one costs people years.

August 24th, two sessions for the University at Buffalo on Zoom. August 25th, on stage at Adobe in San Francisco, the first time I'd given it to a room. Then back to Buffalo the following week for four more in person, opening the semester for the School of Management.

The logistics were bad. Airports, hotels, a talk scheduled to start while my plane was still on the runway at SFO.

But the logistics weren't the cost. The cost was my wife Domenique having the house for eleven days. Mornings, pickups, bedtimes, alone, so I could go stand in rooms I was thrilled to be standing in. I ran the test on that too. Hard, not wrong. There's a real price and somebody at home is paying it, and the only honest response I've got is saying so out loud, often, to her.

---

You can't fix sucks.

Mid-run I was in a hotel room at night working on one slide.

Not the talk. One slide. Moving a line, changing a word, trying to make it land the way I heard it in my head.

Then it registered that I'd already delivered that slide three times in two days and it hadn't worked once.

I wasn't fixing it. I was decorating it.

So I said it out loud: this sucks. And you can't fix sucks. You can only cut it.

I cut it. By the last delivery I wasn't using slides at all.

I'd have kept working on that slide for a month at two talks a year. Not because I'm stubborn, but because at that rate there's no evidence. Three failures in two days is a fact. Three failures across eighteen months is a feeling.

---

The part I want you to take.

Gay Hendricks calls this your Zone of Genius in The Big Leap, and the model has four tiers, not one.

The one that matters is the third. Excellence is where you're very good, well paid, and completely stuck. It's the most comfortable place to spend a career and the hardest to leave, because nothing is wrong.

It's also the tier AI is coming for first. Excellence is, by definition, doing well what plenty of people do adequately.

Hendricks ties genius to doing something better than anyone else, which is the part people drop when they quote him. The trouble is that's hard to measure from inside, because what comes naturally to you doesn't feel special.

So use a cheaper tell. Not whether you're good at it. Whether you finish and want to go again.

---

Two ways to actually do this.

Book the reps before you feel ready. I'm booking my own now. Coffee shops and small rooms in Brooklyn, whatever will have me. And the reps that fixed the talk weren't the impressive ones anyway. Four of the seven were a classroom, two were Zoom, and the Adobe stage was the shortest of the lot.

Find what you keep correcting. Take the work you actually do, for a client or for your job. Then look at what you keep overriding when AI helps you with it.

Here's mine.

AI can write a keynote. It structures an argument, punches up a flat line, cuts a paragraph that's doing nothing. I use it for all of that and it's good at it.

What it can't do is feel the room.

In Hard or Wrong I talk about alcoholism. I talk about losing my son. When I come off that passage the room is carrying something, and every model I've worked with wants to go straight into the next teaching point. On paper that's the correct structural move. The argument continues.

It's wrong. The next slide has to lift, slowly, before anybody can take in another idea. The room needs somewhere to put the thing down first.

I correct that every time. And it isn't a prompting problem. I can't instruct my way out of it, because it takes standing in a room and reading what just happened to the people in it.

That's the talent. Not writing the keynote. Knowing what an audience is carrying out of a passage, and what has to come next.

A speaker coach would catch that. AI doesn't, because it has never watched a room go quiet.

Everything you stopped correcting is everything that just became available to everyone.

And here's what makes it work you can charge for. The thing you keep correcting is the thing everybody else is shipping. They don't know it's wrong. The output looked fine to them, because it does look fine, and you're the only one in the room who can see otherwise.

That's not a preference. That's the job, and staying quiet about it isn't modesty.

That's where Harder to Replace starts. We work out what's still yours, turn it into one offer somebody will pay for, then I put the workflow I use to run growth at EveryExpert on screen so you can build your own. One hour, Wednesday the 16th at noon: save your seat here.

---

One thing this week.

Open the last three things you made with AI for work that matters. Not a draft tweet. Something you shipped.

Read your version against what the model gave you. Every place you changed it, write down what you changed and why, in one line.

Then sort the list in two. Prompting gap if a better instruction would have got you there. Yours if it took looking at the thing and knowing.

Most of the first list will be prompting gaps. That's expected. Those close as you get better at asking.

What's left is short. That's the point. It's supposed to be short, and specific, and it's the part worth money.

Bring it to the 16th and we'll turn it into an offer: save your seat here.

Keep building,
Terry

P.S. Harder to Replace is next Wednesday, 16 September, noon ET. Three things: what you keep correcting AI on and why that's your talent, how to turn it into one offer with one price and one link, and the actual workflow I run at EveryExpert, shown on screen. Save your seat here. Can't make it live? Register anyway and I'll send the recording.

P.P.S. If somebody you work with keeps saying they're worried about AI, forward them this and send them the same link. An hour on the 16th is cheaper than a year of quietly hoping: register here.