
Matt Shumer wrote something last week that broke the internet. If you haven't read it, go read it. I'll wait. Here is the link. Don't forget to come back after you read it.
He compared where we are with AI to February 2020, (you know right before we shut down for Covid) the "this seems overblown" phase right before the world changed. He talked about a managing partner at a major law firm who spends hours every day in AI tools. About how the people refusing to engage will struggle the most. About how the bar is on the floor and almost nobody is doing the work.
I read it and thought: I've been saying this for three years.
Not predicting it. Living it.
In April 2023, I took a $110M software company and went all-in on AI. No consultants. No committees. No 18-month roadmap. Me first. Every Wednesday, mandatory. Four hours building, one hour demos, video on, no hiding.
Shumer says the CEO of a law firm isn't too proud to spend hours a day in AI. I spend 40-50% of my time in these tools. My team averages 62%. We track it. We publish it. Monkey see, monkey do, your team watches what you do, not what you say.
He says "spend one hour a day experimenting." I agree with everything in his essay except that. One hour isn't enough. We do four hours every Wednesday plus demos and I expect my team to AI for 50% of their time working a week. If you add that up, that is 24 hours a week minimum.
He says most people judge AI based on the free version. We invested in enterprise tools from day one. No company data on free tiers. Spending the money will be worth it in the long term.
He says 50% of entry-level white-collar jobs could be gone in one to five years. We are rewriting whole software platforms, one recently went from 3 million lines of code to 125,000. Product managers are deploying features without engineers. 80% of our customer support are resolved by AI agents not humans. Our bug backlog is at a 34-year low. We 4x'd revenue per headcount and we are only getting started.
Shumer wrote a warning letter. I respect it. The world needed to hear it.
But I'm not writing warnings. I'm writing recipes.
Stop learning about AI. Start doing AI. Everything else will fall into place.