
The problem
Legacy code isn't just old — it actively slows you down.
The typical estate is inconsistent, heavily customised, and undocumented, with fragile releases and knowledge that lives in people's heads. Every change is slow and expensive — the cost grows with each new feature.
Little or no automated testing, so every change is risky.
Knowledge lives in people's heads rather than in code or docs.
Fragile, hand-built processes for changing and releasing it.
Inconsistent code and heavy customisation from years of patches.
Estate often grown from acquisitions, each one bringing its own old systems.


What we do
A set of refactoring practices, not a magic machine.
Concrete engineering improvements a team works an application through before it's allowed onto the modern platform. AI itself is one of the main tools used to do the modernising — done in the right order.
Test automation
An automated safety net so anything a human or agent breaks is caught immediately. The single most important prerequisite for letting AI work on the code.
CI/CD
An automated path from a developer's change all the way to production, without slow manual steps.
Packaging & documentation
Standardising how the app is bundled and shipped. Writing down how the system actually works, including in a form agents can use as context.
Security & access
Making sure agents and automation can't reach things they shouldn't, and that credentials and sensitive data are handled correctly.


How it runs
Each app modernised once, then reused.
01
Start
1–2 apps
Start on real applications
Pick one or two apps from the estate. They become the base for the first version of the AI Native factory — the MVP.
02
Bridge
Sandbox
Bridge, then migrate
Build the MVP in a sandbox, gradually connecting it to the legacy estate. It goes live bridged to the old systems, so the two work together during the transition.
03
Compound
Days to weeks
Compound the process
Use what you learned to build a repeatable refactoring and onboarding flow. Each following app takes days to weeks. Retire the old setup once the relevant parts have migrated.
01
Start
1–2 apps
Start on real applications
Pick one or two apps from the estate. They become the base for the first version of the AI Native factory — the MVP.
02
Bridge
Sandbox
Bridge, then migrate
Build the MVP in a sandbox, gradually connecting it to the legacy estate. It goes live bridged to the old systems, so the two work together during the transition.
03
Compound
Days to weeks
Compound the process
Use what you learned to build a repeatable refactoring and onboarding flow. Each following app takes days to weeks. Retire the old setup once the relevant parts have migrated.
01
Start
1–2 apps
Start on real applications
Pick one or two apps from the estate. They become the base for the first version of the AI Native factory — the MVP.
02
Bridge
Sandbox
Bridge, then migrate
Build the MVP in a sandbox, gradually connecting it to the legacy estate. It goes live bridged to the old systems, so the two work together during the transition.
03
Compound
Days to weeks
Compound the process
Use what you learned to build a repeatable refactoring and onboarding flow. Each following app takes days to weeks. Retire the old setup once the relevant parts have migrated.
The through-line
Cloud Native is the foundation. AI Native is the destination.
Reaching Cloud Native — across delivery, architecture, and infrastructure — is the prerequisite for AI Native. The closer you are, the smoother the transition.
Why this approach holds up
One platform, not two
Modernised legacy runs on the same factory as new code. No permanent second world.
Proven method
The same approach re:cinq's leadership ran at Adidas, Shell, and Inditex through cloud native.
20+ years
Two decades of enterprise engineering transformations by the same team.
The book
Cloud Native Transformation (O'Reilly), the leading book on the subject, written by re:cinq's leadership.
AI does the modernising
Once foundations are in place, agents help refactor the rest of the estate. Legacy becomes a source of improvement, not a liability.
