Pragmatic AI. Real results.
Writing code has not been the bottleneck for some time now. With modern AI tooling, it is even less so. The critical path in enterprise software delivery is everything that comes before code: understanding the business domain, modelling how the system should behave, navigating organisational complexity, and describing requirements with enough precision that they can be translated into working software.
Once that thinking is done well, AI-assisted tools make the translation from design to code faster than at any point in the industry's history.
This is not a shortcut - it raises the bar on the work that actually matters: architecture, analysis, and clear communication with stakeholders.
Code generation
GitHub Copilot and Claude Code are part of my daily workflow. I use them to produce production code and test suites at speed - then review, refine, and integrate the output with the same rigour I would apply to any hand-written code.
Testing and Quality
AI-assisted test generation means broader coverage in less time. I use it to generate unit tests, edge-case scenarios, and boilerplate - freeing up time for the test design decisions that require human judgement.
AI tools help me draft technical documentation, architecture decision records, and stakeholder-facing materials faster - so more time goes to thinking and less to formatting.
Documentation and Communication
There is no shortage of AI hype in the industry. My approach is simpler: use AI where it delivers a measurable productivity gain, and skip it where it does not. Every tool I adopt has to earn its place through practical results - faster delivery, better test coverage, or clearer documentation. If it does not improve the outcome,
it does not make it into the workflow.