Buy Harvey, or build your own? A smarter way to decide.
A four-part way to think through any AI tool decision — buy it or build it — plus a two-minute tool to test your own.
Every AI decision you make is a bet on the future. You just haven’t written the bets down.
That’s the idea behind a thought-provoking Every.to article by Dan Pupius, the CTO at The General Partnership. He argues that people building with AI should make their hidden bets explicit — so they notice sooner when one stops being true.
It’s aimed at startup founders. But it maps almost perfectly onto a question a lot of us are wrestling with: do you pay for a legal-specific AI tool like Harvey, or build your own setup with general tools like Claude and ChatGPT?
The four bets you’re making
Pupius breaks the decision into four questions. Here’s how each one lands for a lawyer weighing a specific tool:
What am I really paying for? Is the value smarter answers, or plugging into my systems — my documents, court forms, and client matters? Raw smarts get cheap fast. Integration is the hard-to-copy part.
How soon will a cheap general tool catch up? If Claude or ChatGPT will do the job about as well within a year, don’t lock into a long, pricey contract.
How hard would it be to leave? Can I export my data and walk away, or am I stuck? Favor the bets you can reverse.
Does it protect client confidentiality? No training on my data, matters kept private, and a record of what the tool did.
The bet that flips for lawyers
For a founder, that last one — confidentiality and governance — is a bet you can defer. Move fast now, deal with the rules later.
For us, it’s not optional. Whatever the tool costs and however smart it is, client confidentiality comes first. That single difference changes the whole calculation.
The habit worth stealing
You don’t need the framework to buy this. You need it to think clearly. For any tool you’re weighing, write down four things:
What am I assuming?
What would have to be true for that to pay off?
What would tell me I’m wrong?
How fast could I switch if I am?
Buying can be the right call — especially when you’re paying for real integration and airtight confidentiality. The point isn’t build versus buy. It’s knowing which bet you’re making.
Test your own bets
I built a simple tool that walks you through the four bets. Move four sliders and it shows you which of your bets are safe, which are risky, and whether you lean toward buying or building. It takes about two minutes, and nothing you enter is saved or sent anywhere.
;-)
Ernie
P.S. The tool is the easy part. The harder question is how to score your own bets — and whether your read is right. That’s exactly the kind of thing we work through together in the Inner Circle, so you’re not guessing alone.
Come, compare notes with us
→ https://innercircle.ernietheattorney.net