The quiet death of the SaaS middle

I’ve noticed lately that SaaS is dying. And myself and my team is part of the reason.

A while back someone on my team rebuilt an internal tool that replaced our project management subscription. Not augmented it. Replaced it. Same workflows, same features we actually used, none of the ones we didn’t. It’s live and running today. We’re doing the same thing to at least 3 more SaaS subscriptions costing us monthly.

Two years ago I would have paid a vendor for all of it, or planned out a six-month engineering project.

The cost now? A few days, some tokens, and a team that understands the problem better than any external PM ever could, because it’s our problem.

This isn’t just us. Retool’s 2026 report says that 78% of organizations plan to build more of their own tools this year.

The economics of “buy” assumed building was hard; it isn’t anymore. The economics of “subscribe” assumed your needs were generic; they never have been. The economics of “vendor lock-in” assumed switching costs were high; the costs have dropped significantly.

Not everything dies. Some systems of record, payments, and anything regulated will stay for now. But the middle? The tools that are 80% features you’ve never touched, that layer is shrinking.

This changes what we teach. If making your own version of a tool now takes a weekend, writing the code was never the hard part. Knowing what’s worth building, and what isn’t, is the whole job.