Run
Maintenance
& Support
Building is the cheap part now. Software built in weeks still has to run for years — models are retired, dependencies age, costs drift, and code nobody fully read is the first to break. We take ownership of what is already in production, whether we built it or not.
What's covered
Someone who answers
when it breaks.
Monitoring & incidents
Uptime, errors and performance watched continuously, with agreed response times and a written account of every incident.
Maintenance releases
Bug fixes, OS and browser changes, third-party API changes — shipped through tested releases, not hot fixes on a Friday.
Security reviews & patching
Regular reviews led by our CTO, dependency patching, secrets hygiene and access control — sized for your business, not a bank's.
AI model upgrades
When a provider retires or changes a model, we test the replacement against your evaluation set before anything switches.
AI cost control
Cost per task tracked and the right model used for each job, so usage growth does not turn into a surprise invoice.
Small improvements
The steady stream of small changes a live system needs, without opening a new project for each one.
How it works
Audit, stabilise,
then keep it running.
Handover audit
Code, infrastructure, dependencies, security and documentation — a written picture of what we are taking on.
Stabilise
Fix the urgent risks, add monitoring, backups and the tests that were missing.
Retainer
A monthly scope with agreed response times, regular security reviews and maintenance releases.
Report
A regular summary of what changed, what we found, what it cost to run, and what we recommend next.
Where the cost moved.
When code took months to write, building was the biggest line in the budget. Now a working system can appear in weeks — and the long tail of running it is what costs money: the model that gets retired, the dependency with a security hole, the integration that changes its API, the usage bill that doubles.
That work is unglamorous and it never ends, which is why it needs an owner. We would rather be that owner than the team that built something and left.
Built something yourself with AI tools and not sure it is safe to put in front of users? Start with a production readiness review.
Common questions