I didn't want to ship a golden retriever
Why most AI is too agreeable to actually help you
One quick thing before the story. Next Tuesday, August 4 at 12pm ET, I’m running a free hour on how to be the most memorable person in the room. Live, free, and you’ll leave with a line you can actually use.
For fourteen years I’ve told people that humor is a skill you can build. And for fourteen years, when someone asked me how, my honest answer was they just had to “start.”
Then I wrote a book they could read and later an assessment they could take, but never anything that actually got them “doing” instead of “learning.”
The question shows up after almost every keynote. Someone waits out the line at the front of the room, waits for the handshake photos to finish, and then asks some version of the same thing. “Okay, but how do I actually get better at this?”
For years I gave them a real answer that wasn’t all that useful. Read the book. Take the assessment. Pay attention in meetings. Notice what lands.
All of that is true... but none of it is practice.
The people I’ve watched actually get better at this weren’t the ones who read the most. They were the handful I coached one-on-one. What they had in common wasn’t talent or a good tip. It was that they showed up every week, and somebody told them the truth of what worked and what didn’t.
What took me so long
But there are limits to one-on-one coaching. Cost, availability, and my introvert energy levels.
I’m also just one audience member. I usually have a good sense of what will land, but I can never know for sure.
So the obvious answer was to build something where people could get feedback that didn’t depend on my schedule, or on me suggesting clients sign up for an open mic with a bunch of new comedians making jokes about bodily functions. Something that had the right mix of instruction (so you know what you’re supposed to do), practice (so you actually get to try it out), and feedback (so you know what actually works).
But I never had the right solution because every version I imagined was worse than nothing.
The trouble with AI that gives you feedback on humor is that AI is a people-pleaser. It wants you to feel good. Ask ChatGPT whether your joke is funny and it will tell you it’s wonderful, then offer to write you three more that are somehow worse. Positive encouragement is needed, but false flattery doesn’t make you better. I didn’t want to ship a golden retriever.
What took the time was building a tool that would tell you your second idea was the better idea and why. A coach trained on my material that pushes back instead of applauding as you grow through the eight Humor Intelligence competencies, all in 10 minutes a day.
All of that became the Humor Intelligence app.
Ten weeks, one competency at a time, with a coach named Drewbot who is more honest with you than I would be to your face.
Drewbot handles the daily reps. The part a bot can’t do is still mine, so the founding cohort meets with me live every week for ten weeks, an hour at a time, all recorded for whoever can’t make a Tuesday.
The founding cohort starts Tuesday, August 11.
If you want in
Enrollment is open through Friday, August 7 at 11:59pm PT.
The founding rate is $199.20 a year instead of $249. It’s a 20% discount that stays with you through every renewal, not a first-year teaser. Ten weekly live sessions at 12pm ET, an hour each, all recorded. Seven-day refund if it turns out this isn’t for you.
Come join me on ten weeks of hilarity.
Wit regards,
-Andrew
P.S. If you’re on the fence, I’d rather know why than not. Hit reply with whatever’s holding you back and I’ll answer it straight, no chaser.
P.P.S. If Tuesday’s free hour is more your speed than a ten-week commitment, that’s a completely reasonable place to start. Same link as the top.