For most of the last two decades, the buy vs build decision was simple.
If you needed software, you bought it.
CRM, CMS, CDP, Ecommerce, messaging platforms. Each one came with a predefined model of how your business should work. Not perfect, but close enough. And far easier than building it yourself. So companies adapted. They shaped their processes around the software, not the other way around. That tradeoff made sense because building your own systems was expensive, slow, and risky.
AI changes that constraint.
What used to take teams and months can now be generated in hours. Not perfectly, but well enough to replace real parts of the stack. And that shifts the question. From which software should we buy to whether we should be buying it at all.
Most SaaS products are designed to work for as many companies as possible. That’s their strength. It’s also the limitation. The broader the system, the less precisely it fits any one business. That’s why even great platforms require customization and integration. Not because they’re flawed, but because they’re designed for everyone. For a long time, that was still the better option. Now it’s not always clear that it is.
I’ve seen this firsthand working on multi-tenant marketing platforms. The real value wasn’t the interface. It was the system underneath. The data models, workflows, and how everything fit together across hundreds of locations. The closer the system matched the operating model, the more effective it was. And it rarely matched perfectly.
SaaS didn’t win because it was better than custom software. It won because custom software was too expensive.
AI reopens that decision.
Now companies can build systems that match how they actually operate. Their workflows. Their data. Their constraints. At a certain point, buying software starts to feel like compromise. This doesn’t mean SaaS disappears. It means it moves.
The application layer becomes fluid. Interfaces and workflows can be generated and reshaped as needed. What remains are the hard parts. Identity, data, messaging, and the guarantees that come with them. Which changes where the value is.
It’s no longer in the tool itself, but in the pattern behind it. What actually works. The systems that hold up under real conditions, not just ideal ones. That’s where advantage starts to shift.
To the patterns that define how things should work.
To the infrastructure that makes them reliable.
And to the integration that brings it all together.
For product leaders, the job changes with it.
For years, the question was whether you could afford to build your own systems.
Now it’s whether you can afford not to.
And for the companies selling software, the question changes too.
What’s left that customers can’t build themselves?