Just start with breakfast

BLOCKERS — a weekly series on removing obstacles to success in your practice.

Last week I promised you a video, and I told you I made it with AI.

Here it is.

That’s not me.

Well — it’s me, but I wasn’t actually speaking. I built an AI version of myself so I wouldn’t have to be on camera. At first.

Read on.

I want to tell you exactly how long that took.

Twenty hours

Not twenty minutes. Twenty hours, spread over a few weeks, learning how to manufacture a version of myself that could talk to you so that I wouldn’t have to.

Recording clean sample audio.

Getting the lip sync right.

Fixing a version where I sounded like a hostage, or a robot.

Rewriting the script, because a real person can carry a sentence that a synthetic one can’t.

Re-rendering. Watching it back. Adjusting. Re-rendering again.

Twenty hours.

Sitting down and talking to you like a person took me five minutes.

I know that number is right, because I’ve now done it, and it took about five minutes with my phone and a cheap mic from Amazon that came the same day I ordered it. See the video of the real me on instagram.

What I actually demonstrated

I’ve spent two issues of this newsletter telling you to find the thing you’re avoiding and remove it so that you could build and scale your practice.

Then I spent twenty hours engineering a way to avoid my own blocker — which turned out to be simply being on camera — and I published it and called it a demonstration.

It was a demonstration. Just not the one I meant it to be.

I put that video at the top rather than saving it for the end, because the gap between those two things is the most useful thing I can hand you this week.

You are doing your own version of blocking yourself right now. Maybe not with creating an avatar, but with something YOU are blocked by. Something that takes twenty hours, feels like progress when you do it, produces receipts, and sits directly on top of a five-minute thing you don’t want to do, like paper covers rock.

That thing you don’t want to do is why your practice isn’t growing.

Not the market. Not the economy. Not your budget.

Let me show you how being blocked works from the inside, and then I’ll show you the way out. It’s smaller than you think.

What twenty hours looks like from the outside

Here’s the rest of how I spent the last few weeks.

I read about microphones. I compared a shotgun mic to a dynamic mic. I learned what a boom arm costs. I watched a video about whether an iPhone shoots better than a webcam — it does, and I can now explain why in more detail than you will ever want.

Then framing. Should the microphone be in the shot? There’s a look where the mic hangs into frame and it signals that a real conversation is happening. I thought about that for longer than I want to put in writing.

Then lighting. Then background. Then whether my office reads as professional or as a guy in a room.

Every hour of it felt like work. It looked like work. If you’d walked past my desk, you’d have said: there’s a man preparing.

I wasn’t preparing. I was hiding, in a way that produced receipts. And eventually Claude AI called me out!

I already knew the answer

That’s the part that stings.

The gear doesn’t matter. Nobody has ever watched a video from someone they respect and thought, this would land better with a softer key light. They watch because the person said something true, or they don’t watch at all.

I knew that three weeks ago. I could have written that exact sentence before I spent a dollar or an hour.

Knowing it changed nothing.

Which tells you what kind of problem this is — and it is not the kind you solve with more information.

The moment “somebody” said it to me

You’ll appreciate this, given what last week’s issue was about.

I was working through all of it with Claude AI. And I mean all of it. We were like peanut butter and jelly. I asked about microphones. Claude made the recommendation. Whether to put the mic in the shot. Check. Whether my webcam was good enough or whether I needed the iPhone. Portrait or landscape. Which platform to record on. Backgrounds. And so on.

Claude answered every single question. Patiently. In detail. Genuinely helpful answers, every time.

And I kept asking more. I even ran some of it by ChatGPT. (My wife calls him chatty P)

Then somewhere in there, I asked one more logistics question, and instead of answering it, it said something like:

go shoot it. 🤖

That was it. That was the whole intervention.

I want to be honest about what happened next, because it isn’t heroic. I sat there for a second feeling mildly called out by a piece of software. I felt like someone real was in my office.

Then I picked up my phone, and I recorded the thing, and it took five minutes. See it on instagram.

Here’s what I’ve been chewing on since.

The AI wasn’t smarter than me about video. It didn’t know something I didn’t. It just wasn’t invested in my excuses, and it had no reason to keep helping me research a problem I’d already solved twenty hours earlier.

That’s the actual value, and it’s the thing I was pointing at last week without quite seeing it. Not the answers. Somebody — something — willing to notice that you’re asking the wrong question. I can help you with that.

Just start with breakfast

You already know how this works. You do it every week.

Think about the patient with the new diagnosis who has read everything. She can explain her own condition better than half the residents. She knows the mechanism, the guidelines, the interventions, the studies. And she has changed exactly nothing about how she eats.

You don’t send her more information. She is drowning in information.

What she’s missing is one action small enough to survive contact with an actual Tuesday.

So you don’t say “overhaul your diet.” You say: this week, breakfast. That’s all. Just breakfast.

You have never once believed that woman’s problem was knowledge.

And yet there I was, twenty hours deep, building a robot to do my talking for me.

I didn’t need a better camera, or a better avatar. I needed my breakfast — the one thing small enough that I’d actually do it.

Mine was five minutes and a phone. Try it.

What to do today

Grab your phone. Right now, where you’re sitting. Answer one question a patient asked you this week. Don’t script it. Explain leaky gut or hormone imbalance. Two takes.

Do not watch it back today. That’s the trap. Today you’d only be hunting for flaws. Watch it tomorrow and you’ll see a person explaining something clearly.

Publish the second one, not the first. The first one exists to get the flinch out of your system. Nobody has to see it.

If you catch yourself opening a tab to research which app to use — that’s the twenty hours starting. Close it. I already did that for you 😉

No purchase required. The phone in your hand is better than the equipment most of the internet was built on.

And if you want the version of this that worked for me: ask an AI what you should do first, and then do the thing it says instead of asking it a follow-up question.

And where I come in

Nothing here needs me. I’d rather you record ninety seconds today than book a call with me about it.

But I’ll say the part that connects to what I actually sell, because it’s true and it took me a while to see it.

The video isn’t the deliverable. It’s one piece of what your patients experience when they’re not sitting in front of you. So is the program. So is the check-in on day nine. So is the lab reminder, and the article you never wrote.

It’s all the same category — what happens between visits, when you’re not in the room.

Most practices have nothing there at all. That’s a much bigger blocker than any of us enjoy looking at, and it’s the one I’m genuinely useful for.

But it starts with ninety seconds on a phone.

Go hit record.

Or email me and say blocked, and I’ll help you find your first thing.

John Cummings
Founder, BodySite.com

P.S. — I’m still going to buy a better microphone. Just after I record, not before. Which is the only order that has ever worked for anything.

Catch up on the series: Episode 1 — Stop looking at the car · Episode 2 — What should I do first?