When your production system is one deploy from an incident.
I’m Harry Boyes. For fifteen years I’ve been called into fragile Rails, C# and Node codebases to stop the bleeding, then modernise them incrementally, without the rewrites that usually make things worse.
Fragile software quietly becomes a business problem.
Every release feels risky. Small changes cause outsized incidents. Your best engineers spend more time firefighting than shipping. The security audit is coming and nobody wants to open the CVE list.
The instinct is to rewrite. It’s almost always the wrong move. Rewrites take twice as long as planned, ship without the edge cases the old system quietly handled, and leave the business flying blind during the transition.
There is a calmer, cheaper path: stabilise first, then modernise in place, one seam at a time.
Incremental. Risk-aware. Boring on purpose.
-
01
Stabilise
Fix the loudest reliability, security and performance issues so the system stops burning attention.
-
02
Understand
Map the actual behaviour of the system, not the intended behaviour. Document seams, coupling and risk.
-
03
Contain
Add the tests, telemetry and safety nets that make change reversible. Nothing modernises without them.
-
04
Modernise
Replace old code in small, verifiable steps using strangler-fig patterns. Never a big-bang cutover.
Matalan. A 36-hour migration, in 30 minutes.
Millions of images. A one-shot upload window with no rollback. I rebuilt it as a distributed ETL pipeline with idempotent checkpoints, running in parallel across small services.
Read the full write-up →Three engagements. Start wherever the risk is highest.
Legacy system triage call
A 60 to 90 minute call. Walk through the system, the pain, and whether it’s worth fixing. You leave with a clear read on risk and options.
Book a call →Assessment & migration plan
Two to four weeks. I get properly inside the system and produce a pragmatic plan to modernise it, with cost, sequencing and risk called out.
See what's included →Migration & modernisation
Two to six months, longer if it needs it. I take the problem off your plate and ship a system you can trust again.
See engagement details →Two other things I build for a living.
SaaS builds & MVPs
One senior engineer, embedded, shipping every week. From idea to a working product in weeks, not quarters.
SEE HOW I BUILD →Small-business websites
A proper website for your business. Fixed price, hosting sorted, built by a real person in Nottingham.
SEE THE PACKAGE →Recent essays.
The Price of Fragile Systems
Legacy software doesn't just make development harder. It limits how a business can respond to change.
READ ESSAY →Legacy software doesn't stop work, it just creates risk
When software gets in the way, people find workarounds. Those workarounds introduce risk.
READ ESSAY →