Welcome back to the Launch Key 🚀

There are countless entrepreneurial articles about experimentation. I’ve been very lucky applying that concept to technology.

The AI panel discussed it at last week’s CEO Convergence too. And on my 3-hour drive back to the mountains I had better answers than I gave in person. Some of the things I should have said became this week’s Launch Key.

TLDR: there is a dividend for tinkering.

Let's get into it.

"Are we missing the boat?"

That was the question underneath many questions at CEO Convergence last week — a one-day gathering of Anything but Typical business owners led by Gary Frey and Ben McDonald. I was on the AI panel.

These were people who've built real companies. Made payroll. Survived a recession or two. And they were worried they'd already fallen behind.

Here's what I told them.

I quoted Marc Andreessen: we're roughly three years into this. (On stage I said a thirty-year AI revolution. He actually frames it as eighty. Which makes the point even better.)

If we're that early, the most important AI companies probably haven't been invented yet. Neither have the most important uses of AI inside your company.

Then came a follow-up: what if we did nothing for twelve months?

In reality — you'd probably be fine.

But fine isn't the goal. The risk isn't falling behind. It's missing the dividend.

Tim Denning put it bluntly in a recent post: consuming more information is a form of procrastination. The smartest move is to use what you already know and go experiment.

Webinars. White papers. Newsletters. (Yes, I see the irony!) At some point, reading about AI becomes a very sophisticated way of not doing anything with it.

The people in that room who were getting somewhere had one thing in common. They just do stuff. They've succeeded and they've failed - both because they tried.

Nobody handed them an AI strategy. Nobody's going to hand you one either.

Denning's line for this is "write your own damn permission slip."

What’s holding you back?

I've written myself that slip half a dozen times in forty years. I never once picked the winner on purpose.

Take HyperCard. Apple shipped it free with every Mac in 1987. Non-technical people could just make things with Hypercard. I didn’t have the talent to create Myst, but instead used it to digitize files at my headhunting job. And then build myself a GMAT prep guide — nothing grander than trying to pass a test.

On my first day of grad school at the University of Georgia, Dr. Scott Shamp asked everyone to name something interesting they were doing outside of class. I mentioned the guide.

"See me after class."

It turned out that Shamp was using HyperCard to drive laserdisc multimedia projects and needed research assistants. I spent two years building with him — including a project on the Challenger explosion. It set my direction for decades in technology.

I didn't learn HyperCard to get into multimedia. I learned it to organize my files, build a little database and then pass a test. The door showed up anyway.

That's been the pattern ever since:

The payoff was never the thing I set out to learn. It was the door the experiment opened.

And every one of these looked like tinkering at the start. That's exactly when it's cheapest.

Here's the recipe I gave the room. It works for a fifty-person company, and it works for a business of one.

Pick one small, tactical thing that happens all the time — the repetitive, every-week stuff. Then check four boxes:

Now ask: can I build an AI experiment that takes this off someone's plate?

That's it. Not a transformation. Not a roadmap. One task.

My friend Patrick McLean said it better than I can: everybody starts too big.

Someone asked which AI model to use. I gave a generic "you need to test different models" answer. True, but not useful.

Here's what I should have said.

Last year I experimented with ChatGPT, Jasper, Writer and Perplexity before settling into Claude in early 2026. Then when SpaceX bought Cursor, I thought Elon Musk would have a more serious AI stack - so I experimented with Grok. Now I have created dueling research agents working for Launch Key — one built on Claude, one on Grok. They bring me their findings. I grade the work and decide what’s worth keeping before any of it gets added to the pipeline for this newsletter.

That's the whole game. The models will keep changing — every few months there's a new leader. Your business problems won't. Keep your eye on the problems, swap the tools, and grade the output like you'd grade any new hire.

Someone else asked what all this looks like in thirty years. I didn't answer that one well either, because I don't know. Neither does anyone who tells you they do.

What I do know is what happens when you tinker. Forty years of it. The doors keep showing up.

Back in May I wrote about building your own door. This is the same idea with a tool in your hand.

So here's your assignment for the week: pick your one task. The one that passes all four checks. Build a small experiment.

Hit reply if you care to share. I read these, and the best ones may show up in a future issue.

Write your own permission slip.

Now go launch something 🚀

Thousands of AI papers, model releases, and product launches ship every week. You don't need all of them. You need the ten that matter.

TLDR AI is the free daily newsletter curated by Anthropic and ex-Google engineers. People who build AI for a living read everything, then send you what's actually worth your time: the models, research, and tools that will still matter next month.

No hot takes, no hype cycles, no 60-tweet threads. Clear summaries you can read between meetings.

Join 1.1M+ readers. Subscribe for free.

Subscribe for free

It’s so difficult to keep up with the pace of AI, that I built my own daily briefing agent. But who needs another email? (see Eat Your Own Dog Food)

So I had Claude leverage design skills to create an HTML Futurist page delivered daily at 5am to Github. Markets, News, Crypto, a pull quote and a little daily commentary from a techno optimist (think, Marc Andreessen voice).

It’s just an aggregation play and sources (RSS, X, YouTube) are listed in the footer. I wholeheartedly ask for your opinions. Do you find this useful? Who else should be included for insight? Comment below or email suggestions.

📕 MGMT Playbook : Practical management insights straight to your inbox every Wednesday.

🏢 Retirepreneur : Your guide to turning decades of experience into profitable consulting, coaching, or course businesses, without risking your retirement security.

🗃 Dealroom Business Success Uncovered : Learn directly from billionaire entrepreneurs on how to grow a business. Join a community of 2,000+ innovators.

🚏 The Cautionary : Learn to challenge conventional wisdom.

🧾 Redefining Retirement : Too Young to Retire, Too Old To Take Orders? Yeah, Me Too.