Why Do We Keep Building the Wrong AI Stuff?
April 26, 2026 In the February 1st newsletter I talked about how AI agents had started their own language, formed a cult, and created online communities that explicitly forbade humans from joining. I concluded with a meme from Jurassic Park (the original, not the crappy knockoffs from the last decade).
I can’t believe I’m going back to the Jurassic Park well again, but when I was trying to articulate a thought, I Googled a phrase and it turned out to be another Dr Ian Malcolm quote.
“Your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should.”
Our teams have been off to the races for a few months now and the results are mind blowing. But I’m also discovering that I need to do a better job of putting up guardrails. They are falling down rabbit holes they don’t need to fall down.
Just because we can build doesn’t mean we should build it.
In fact, it feels like we are focused on building the wrong things. For whatever reason, we seem to build the commoditized part.
This isn’t a setup. I’m not about to give you the answer, because I don’t know it. But there is a definitive pattern I am calling out.
If I think of our secret sauce, I want AI to tap into that. The 20% of the 80/20 split. Last mile. Whatever you want to call it. We all have things that are distinctly unique to us. Let’s leverage AI to amplify that.
But inevitably, we build the 80% part. The commoditized part. I can’t figure it out.
A good buddy of mine recently left a prominent retail shop and started his own thing. To protect the innocent, I will change the spelling of his name to Bryan with a “y”.
His team is small and lean and he was telling me about how he’s trying to build out a property accounting system using AI so he didn’t have to spend money on Yardi. I peppered him. Is that your secret sauce? Is that why investors trust you?
You’ve always had the choice to build out your own property accounting system. It’s not like Yardi had some proprietary formula - they just perfected a process over time. AI might make the effort to build feel doable, but do you really think you will build something better than Yardi? Think about all the edge cases they have uncovered over the years. All the things that broke their software temporarily.
In an attempt to “save” $X, I bet he will spend far more trying to build this out.
It’s impossible to quantify opportunity cost, but I think that that’s the nasty part here. He’s a gifted real estate investor, not an accounting tech guy. What should he be focused on?
But while I am lecturing him, my team is doing the same thing. First thing they brought to me? A CRM. HubSpot drives me nuts, but is it our secret sauce? No, it’s the commoditized part. The worst part? They needed the data from HubSpot to make ours work!
Geoffrey Moore is an organizational theorist and he wrote a book called Dealing with Darwin that outlines a framework for how companies should tackle this problem. I initially thought of this as “buy vs build”, but that’s really the outcome. He tries to outline the decision making process:
Core = what customers pay you for. The work that differentiates you, builds the moat, and compounds over time. If you do it badly, you lose.
Context = everything else you have to do to run the business. Necessary, but nobody chose you because of it. If you do it well, no one notices.
The rule: invest in Core, outsource Context.
My team (and Bryan definitely not with an “i”) are falling into a trap. AI makes Context feel cheap to build, so teams build it.
But cheap to build isn't the same as worth building.
Every hour spent on Context is an hour not spent widening the moat.
So my new rule for the team is ripped right from Moore’s book: if a vendor already does this as their entire business, it's Context. Let someone else do it and pay them for it. That’s the commoditized part. We aren’t going to out Yardi Yardi.
Our AI efforts need to focus exclusively on the Core. Our secret sauce. The differentiating factor. What customers pay us for.
