Replace the old system without stopping the business.
We move it one piece at a time, so your team keeps working while the software changes.
Four parts of a business system moving one at a time from the old system to the new one, until the old system is retired.
Overview
Some systems still run the business but have become hard to change, hard to staff or risky to keep. We start by mapping what the system does and what depends on it, including the rules nobody wrote down. Then we replace it one piece at a time, with the new part running next to the old until the results match.
A good fit if
- It runs on a version past end of support, such as .NET 8 (support ended 10 November 2026) or PHP 8.2 (ends 31 December 2026).
- One person understands how it works.
- Small changes take weeks.
- You pay more each year for extended support.
Not the right start if
- The system is stable, supported and rarely changed. Keep it. We'll say so on the call.
- Only one connection to it is the pain. Start with Custom Software & Integrations
- You want it rewritten from scratch in one go. We replace systems in stages. A big-bang rewrite is the riskiest way to do it.
Cutover
The old system shrinks one piece at a time
A router sends each part of the work to the old system or the new one. Move parts across, roll one back, and switch the old system off when nothing depends on it.
Old system handles: 100%75%50%25%0%
Old system
New system
- Quotes
- Invoices
- Customer records
- Reports
Old system switched off. The router and the bridge code are removed, and the licenses and servers are cancelled.
Everything still runs on the old system. Quotes now run on the new system. Both ran side by side until the results matched. Old system handles: 75%. Invoices are back on the old system in one step. Nothing was lost; the new version gets fixed and moved again. Old system switched off. The router and the bridge code are removed, and the licenses and servers are cancelled. Everything still runs on the old system. now run on the new system. Both ran side by side until the results matched. Old system handles: %. are back on the old system in one step. Nothing was lost; the new version gets fixed and moved again. Old system switched off. The router and the bridge code are removed, and the licenses and servers are cancelled.Every case
- Start
- Everything still runs on the old system.
- After a move
- Quotes now run on the new system. Both ran side by side until the results matched. Old system handles: 75%.
- After a rollback
- Invoices are back on the old system in one step. Nothing was lost; the new version gets fixed and moved again.
- Old system switched off
- Old system switched off. The router and the bridge code are removed, and the licenses and servers are cancelled.
Process
How it runs
You keep
- 01A dependency map and an inventory.
- 02A business rule catalogue and a plain-English spec per module.
- 03The behavior tests.
- 04The router, with a removal date.
- 05A match report for each part.
- 06A cutover runbook and a rollback record.
- 07A decommission record with the savings.
-
01 of 07
Map what exists.
Every screen, job, table and connection, read with AI help and checked by a person.
You keep A dependency map and an inventory.
-
02 of 07
Write down the rules.
What the system actually does, including the rules nobody documented.
You keep A business rule catalogue and a plain-English spec per module.
-
Repeats for each part · Steps 3–6
03 of 07
Pin the behavior.
Tests that record what the old system does today, which the new one must pass too.
You keep The behavior tests.
-
04 of 07
Put a router in front.
One place that decides which system handles each request, marked as temporary.
You keep The router, with a removal date.
-
05 of 07
Move one part.
The new version behind a switch, running beside the old, with daily checks that both agree.
You keep A match report for each part.
-
06 of 07
Switch over, with a way back.
Staff first, then a share of users, then everyone, with a rehearsed rollback.
You keep A cutover runbook and a rollback record.
-
Once · Step 7
07 of 07
Switch the old one off.
Data archived, licenses cancelled, servers shut down, bridge code deleted.
You keep A decommission record with the savings.
Risks
Finished means the old system is off
-
The big rewrite that never lands.
A year goes by and the old system still runs everything.
Prevented in step 5 -
Rules nobody wrote down.
The new system misses behavior the business relied on.
Prevented in step 2 -
The permanent seam.
The last part never moves, so you pay for two systems.
Prevented in step 7
3 of 10
Of the 10 critical federal systems GAO flagged for modernization in 2019, 3 were finished by February 2025.
23 to 60 years
The age of the 11 most critical federal legacy systems in the same audit.
Deliverables
What you keep
-
System assessment
What the system does, what depends on it and where the risk sits.
-
Modernization plan
The order to replace things in, starting where the risk or the cost is highest.
-
Data migration
Your data moved and checked against the old system before anyone relies on it.
-
Staged cutover
One part switched over at a time, with a way back.
-
Old system retired
Switched off once nothing depends on it, with the new system documented.
Done means
- Every part runs on the new system.
- Old and new matched for the agreed period before each switch.
- Old data is archived and still searchable.
- Licences and hosting for the old system are cancelled, with the savings recorded.
- The bridge code is deleted.
- The rule catalogue and behavior tests are handed over as living documentation.
Sample
Business rule catalogue
| Rule 14 | Orders over 500 units get a manual credit check |
|---|---|
| Found in | ORDPROC.cbl, line 812; not in any document |
| Owner | Finance lead |
| Still wanted? | Yes, threshold raised to 1,000 |
| Match report, Invoices | 12,408 of 12,408 matched over 30 days |
FAQ
Questions about Legacy Modernization
Still deciding?
Ask the people who would build it. We reply within one business day.
In stages. The new piece runs next to the old one, we compare results, and we switch over one workflow at a time with a way back.
Very little for each part, because old and new run side by side. The final data move may need a short planned window, which we agree with you in advance.
It depends on what the assessment finds. Sometimes repair costs less than a rewrite, and sometimes the reverse. We say which with reasons, you choose, and the findings are yours either way.
Cheaper to understand the old system, not free to replace it. AI reads old code and drafts specs quickly, but testing, moving data and switching over are still most of the work, and people own those.
Often, for a while. It tends to get dearer: Windows 10 extended updates for businesses cost 61 US dollars per device in the first year and double each year after.
Not to begin. We start with the code and a staging copy, and ask for production access only when the work needs it, with read-only access first.
When the old system is switched off, not when the new one launches.
Contact
Have an idea or a stuck project? Let's talk.
We reply within one business day. If we're not the right team, we'll say so on the first call.
On the first call
- What the system does, and what depends on it.
- Why now: a deadline, a cost or a risk.
- Who still knows how it works.