Build vs. Buy Is Finally Bending

For years my advice on build versus buy was simple: you are probably going to build it, and you are probably wrong to.

That rule earned its keep across my whole operating career. Now, for the first time, it is bending, and if you have not rethought where you draw the line since AI-assisted building got real, you are drawing it from memory.

Why teams always defaulted to build

Product companies build. It is in the DNA. The company exists because it built something, so building feels natural, and every internal problem starts to look like a product problem the team is uniquely qualified to solve.

The math usually said otherwise. A specialized vendor whose entire business is that one problem has more reps at it than your team will ever get, ships improvements you do not have to fund, and amortizes their engineering across hundreds of customers. Your team building the same thing on the side produces a worse version, later, with a permanent maintenance tail that nobody budgets. The build decision is made once, with optimism. The maintenance is paid forever, with interest.

I gave the buy speech many times across my Nav years, and I was right most of the times I gave it.

What changed

AI is collapsing the cost of building. Not shaving it. Collapsing it.

I have watched people with no engineering background spin up real, working applications in a day. I have built my own tools this way at home and in my business: dashboards, forecasting models, commission calculators, workflow automations that would have been a vendor evaluation or a contractor engagement two years ago. Things that used to clearly justify a vendor are suddenly a weekend project, and the weekend project version is customized to exactly how my business runs.

The cost structure that made “buy” the smart default has shifted on one side of the equation. When building something took a quarter and a team, the vendor’s amortized engineering always won. When building it takes an operator and an afternoon, a real category of tools stops justifying a subscription.

Where the line actually moved

Here is how I draw it now, having lived on both sides.

Internal tools with one user or one team have moved hardest toward build. Reporting, internal calculators, glue between systems, personal workflow automation. The failure cost is low, the customization payoff is high, and maintenance is a prompt away. This is where I build without hesitation.

Anything customer-facing still leans buy, or at least leans slow. The bar is uptime, security, and edge cases, and AI-assisted speed does not lower that bar, it just makes it easier to reach the start line. A tool that breaks for me wastes my morning. A tool that breaks for customers spends trust.

Anything regulated, or holding sensitive data, stays buy-biased. Vendors in those categories are selling compliance and liability absorption as much as software, and an afternoon of building does not replicate either.

And the deepest version of buy, the vendor whose product is their entire company and moat, is as defensible as ever. AI did not make it a weekend project to replicate Stripe. It made it a weekend project to replicate the thin SaaS tools that were charging vendor prices for spreadsheet logic. That middle tier is where the reckoning is happening.

What I would do this quarter

Re-run the build-versus-buy analysis on your stack as if you were assembling it today. Not to rip everything out, but because the answer has changed for some line items and inertia is the only thing maintaining them.

Look hardest at the tools your team uses internally that charge per seat for logic your operators could now describe to an agent. Look at the contractor engagements that exist because building used to require engineers. And keep paying, happily, for the vendors that absorb risk, hold the compliance burden, or run something your customers touch.

Buy is still right more often than product teams admit. But the line is moving, and it is moving fast. Worth rethinking where you drew it a year ago, because a year ago is a different technology era now.

For what operator-built tooling looks like in practice, read the owner-operator case study, or get in touch.