Approach
Replace the engine
while the car is moving.
Rewrites fail when they are one enormous bet settled on a single weekend. We break the bet into small ones, each of which can be won, measured, and — if it needs to be — undone.
-
01
Map
We sit with the people who use the system and read the code that runs it. Every business rule buried in a form event handler gets written down.
OutputA working map of the system and what it must never stop doing.
-
02
Prove
One real slice — the process that hurts most — built and put in front of the people who do that job every day.
OutputSomething running you can click, inside weeks.
-
03
Cut over
Module by module, with the old system live alongside. Data migrated with a preview and a way back.
OutputA move nobody had to work a weekend for.
-
04
Compound
Now that change is cheap, AI and automation go where they pay. Then the next thing.
OutputA platform that gets better instead of older.
Non-negotiables
The rules we do not bend.
The old system keeps trading
Until the new one has earned the work, both run. Nothing is switched off because a project plan said this was the week.
Every migration is previewable
You see exactly what will land before it lands, and there is a way back. Data moves are the highest-risk hour of any project — so we make them dull.
The business rules survive
Twenty years of exceptions are in that code because a customer asked for them. We find them, write them down, and confirm each one still applies.
Your data is yours
All of it, exportable in full whenever you ask, in a format something else can read. We hold the platform up; we never hold the data hostage.
AI must be able to say "I don't know"
A confident wrong answer is worse than no feature. Grounded sources, cited, and silence when the sources do not cover it.
Documentation ships with the code
Not afterwards, not in a follow-up. Documentation nobody edits is documentation that lies.
Questions we get asked
Straight answers.
We can't stop trading for a rewrite. How does this work?
It doesn't stop. The old system keeps running while we replace it a module at a time, with both live together and data flowing between them. Cutover is a series of small, reversible steps — not a weekend everyone dreads.
How old is too old?
Nothing is too old. VB6, Access, Delphi, FoxPro, classic ASP, PHP 5, a Windows service nobody has rebuilt since 2011 — these are normal here. If it still runs, we can read it, and what it does is worth more than how it was written.
Where does AI actually fit?
Where it earns its place: answering questions from your own documentation, reading the documents that arrive as email attachments, and taking the guesswork out of estimating and scheduling. It is built into the system people already use, it shows its working, and it is told to say when it doesn't know rather than invent an answer.
Who owns what?
The data is yours — all of it, exportable in full whenever you ask, in a format something else can read. The platform is built for your business, and we build, host and maintain it: no per-seat licence, and what gets built next is decided by your business rather than by somebody else's product roadmap. The commercial terms are written down before any work starts, so none of this is a surprise later.
What does it cost to find out?
A conversation costs nothing. The first paid piece is usually a short discovery: we map the system, write down what it must never stop doing, and come back with a plan, a sequence and a number. You can take that plan anywhere.
How quickly do we see something real?
Weeks, not quarters. The first slice is a working piece of the process that hurts most, running with your data, in front of the people who do that job.
Next step
Tell us what the old system
will not let you do.
Thirty minutes, no pitch deck. You describe the system and what it is costing you; we tell you honestly whether it is worth replacing, and what that would look like.