A DECISION FRAMEWORK
Keep. Fix. Rebuild.
AI can rewrite your legacy code faster than ever. It can't tell you whether you should.
A structured team assessment for deciding what to do with the systems that hold your business back. It weighs technical health, business needs, cost and readiness, and brings every role that knows the system to the table. The result is a decision to fix, refactor in stages or rebuild, based on what everyone knows rather than the loudest voice in the room.
What's in the kit
Free. Two PDFs. Just your email.
The problem
Why "just use AI" isn't the answer
AI sees the code, not the business.
It can't tell you whether the system needs to change in the next two years, what budget is available, or what compliance risk you carry.
AI doesn't know your company.
Hidden behaviour, odd customer requests and workarounds built up over the years often live only in people's heads.
AI can't settle a disagreement.
When the engineers say "rebuild" and product says "it's fine", the answer comes from understanding why they see it differently, not from generating more code.
A faster rebuild is still the wrong call if you shouldn't rebuild.
AI lowers the cost of building. It doesn't make the decision for you.
Where it applies
What the framework can be used for
Use it whenever a system causes enough pain that someone has asked whether it should be replaced.
Legacy code
An older codebase that is slow and risky to change, where a rewrite keeps coming up.
Problem modules and APIs
One part of a larger system, or a fragile, poorly documented API, that causes most of the incidents or is hard to change without breaking what depends on it.
AI rewrite proposals
When someone suggests rewriting with AI, check first whether a rewrite is the right call.
Ageing tools and platforms
Internal tools, vendor products or platforms nearing end of support, or no longer fitting how the business works.
Budget and roadmap planning
Bring a recommendation backed by the whole team into annual planning or a funding request.
A debate that has stalled
When the team has argued fix or rebuild for months without reaching a decision.
Who takes part
Every perspective, one picture
Rebuild decisions often go wrong because each person sees a different part of the picture. This assessment puts every view side by side, so the team can see where it agrees, where it doesn't, and what that means for the decision.
Engineering manager
Team capacity, delivery risk, onboarding time, and whether a rebuild can run alongside daily work.
Engineers
Incidents, slow changes, risky releases, and code only one person understands.
Architect
Outdated technology, hidden behaviour, and what it really takes to move data and users safely.
Product manager
Blocked features, where the business is heading, and whether the cost is worth the value.
And anyone else who knows the system
The kit is open to product owners, scrum masters, QA, operations, security, data or anyone else with a real view of the system. Invite four to ten people who know the system well.
For leadership
Leadership gets what it needs to back the decision: a one-page summary with the reasons, the risks, the cost picture and the ask.
THE PROCESS
How to run it
Choose who takes part
Everyone with a real view of the system.
Score privately
Each person fills in the handout before any discussion, so answers stay honest.
Collect and compare
Average each question and section, and note where scores differ most.
Discuss the disagreements
Start where scores differ by two points or more, then look at cost, then agree the decision, one owner and one next step.
The assessment
Twenty questions, five areas
{{ area.name }}
{{ area.count }} questions{{ area.topics }}
How to read the scores. In A to C and E, a higher score means more pressure to replace. In D, a higher score means more readiness to rebuild.
The outcome
Three possible answers
Fix
Low pressure in A to C
Keep maintaining the system.
Refactor in stages
High pressure in A to C, low readiness in D
Improve step by step and build readiness first.
Rebuild
High pressure in A to C, high readiness in D
And only when the cost check in E holds up.
The role of AI
AI where it helps
Let AI do the reading and the arithmetic, and keep the deciding with your team.
The kit includes a context block and three ready-to-paste prompts: upload the filled-in handouts to Claude or another AI assistant to read the scores, prepare the discussion and draft the leadership summary.
Free download
Download the kit
Everything you need to run the session with your team.
Enter your email to get both documents. You'll be added to the Keep. Fix. Rebuild. mailing list and can unsubscribe at any time.
PDF, 5 pages
Facilitator instructions
How to run the session, read the results, the AI prompts and a CSV fallback template.
Fillable PDF, 6 pages
Participant handout
The 20 questions, to fill in on screen or print.
From the creator
Micquella Galea Scerri
Micquella Galea Scerri is a product manager based in Switzerland. Over nine years in regulated MedTech, she has led digital transformation across enterprise platforms, e-commerce platforms, product information management and digital experience, working with many development teams worldwide. Keep. Fix. Rebuild. grew out of that work.
micquella.galeascerri.comFeedback
Tried it with your team?
A fuller version is in development, shaped by feedback from the teams who use this one. How did it go, and what would have made the decision easier?
micquella@galeascerri.com