You start by wiring together what’s already there. Then it stops being enough, and that’s where the risk is.
The problem
Almost nobody starts from zero. You start by gluing together external services, and that is the right call while the company is small: it’s cheap and it’s fast.
Then the company grows and that arrangement stops answering. Not all at once — one piece at a time, and every piece that gives way is a piece somebody is working on.
The constraint
You can’t stop to rebuild. There are already clients, already professionals opening that software in the morning, already commitments. A full rewrite isn’t a plan: it’s a bet.
How I do it
One piece at a time, in the order that matters. First what can’t stop, then the rest once the first part is safe.
It’s the same move on a system from 2010 and on a startup from 2025 — fifteen years apart, and the move doesn’t change: never replace everything at once.
The proof
A healthcare platform that started exactly that way, from a stack of external services wired together. Today 150 to 200 professionals use it, and it has never broken. The infrastructure went the same way: the third-party video service replaced by a self-hosted one, without interrupting anything.
On a recent product, fourteen releases in forty-seven days.
The first step is a conversation, not a quote.
Tell me what the problem is. If I can solve it I’ll say so, and if I can’t I’ll tell you who can.