The oldest question in software is build or buy. For most small and traditional businesses the answer was always buy, and not because the bought tool fit. It was because building meant hiring developers, managing a project, and carrying a budget that only made sense at scale. So businesses bent themselves around generic software and absorbed the manual work in the gaps.
That calculation just changed.
AI has collapsed the cost of building bespoke software. A single capable operator can now build, in days, tools that used to take a small team and several months: a data pipeline, an internal dashboard, an assistant that handles a specific job, a model that turns messy inputs into a clear call.
When the cost of building falls far enough, the whole build-or-buy decision flips. The bespoke tool that was once an indulgence is now the practical option, because it fits the actual problem instead of the average one.
The interesting consequence is not that big companies get more tools. It is that small ones finally get any. A business that could never justify a custom system can now have one shaped exactly to how it works. The generic tool, and the workarounds it forced, can be retired.
This matters most for traditional businesses. They tend to have two things in abundance: deep knowledge of how their industry actually runs, and very little budget for software. The old maths shut them out. The new maths fits them precisely, because the expensive part, the build, is no longer expensive.
None of this makes AI the point. AI is the how, not the pitch. A tool still has to be aimed at a real gap, a genuine inefficiency or blind spot that costs the business money. Pointed at nothing, AI is theatre: an impressive demo that changes no number.
So the discipline does not change, it gets cheaper to apply. Work from first principles. Find where the business is leaking. Then, and only then, build the fix. What is new is that far more of those fixes are now worth building.
This is the shift I have built Pathway Prosper around. The method was always the same: find the gap, engineer the fix, measure it against the bottom line. AI did not replace that method. It widened the set of businesses the method can serve, and made bespoke an answer almost anyone can afford.
It probably is not anymore. Tell me the gap.