Write the behaviour down first.
Before implementation, I turn the intended behaviour into a clear example or failing test. That gives the change a contract and makes the edge cases visible early.
How we workTDD / documented delivery
A practical way to build with AI
Build it.
Prove it.
I use test-driven development, automated checks, and clear documentation to turn fast-moving ideas into software that can be understood, trusted, and changed.
Start a conversationFast is useful.
Proven is better.
AI can accelerate a first pass, but it does not replace a development method. I start with expected behaviour, build against that contract, run automated checks, and document the decisions that make the system easier to carry forward.
The approachThree useful habits
Make the smallest useful change.
I use AI and existing tools to explore and implement quickly, while keeping the code readable and the change focused on the behaviour the test describes.
Leave evidence that it works.
Automated tests, useful documentation, and a repeatable verification path make the next change safer, whether I am building from scratch or repairing vibe-coded software.
Where I can helpPractical capability
Test-driven development
Define expected behaviour in a test before implementation, then use the feedback loop to keep the change precise and the code honest.
Automated testing
Build repeatable checks around the important paths so regressions are caught by the project, not discovered by the next person using it.
Documented decisions
Record the useful context, assumptions, setup, and trade-offs so the system can be understood beyond the original build conversation.
AI-assisted delivery
Use AI to speed up exploration, implementation, documentation, and repetitive work, while keeping review and responsibility human.
Vibe-code repair
Untangle generated or partially working code, reproduce the failure, add coverage for it, and replace brittle patches with understandable behaviour.
Integration thinking
Connect Shopify, Recharge, Laravel, APIs, webhooks, and third-party services with clear ownership and tested failure handling.
A good fit for this work
Something already exists, but it is not quite right.
You may have a prototype, an AI-generated feature, a stack of connected tools, or a workflow that has grown around its gaps. I can turn the intended behaviour into tests, document what the system is doing, and make the next change safer to deliver.
Test-driven developmentAutomated checksDocumented decisionsExisting codebasesAI-assisted buildsAPI integrations
The standardUseful after launch
Leave the system
better understood.
Good delivery includes a test suite that can be run again, documentation that explains the important decisions, and a short explanation of how the pieces fit together. The aim is not just to ship the feature, but to make the next change safer.