AI training for developers
We work through a software task from gathering context to testing and reviewing the changes. Participants learn where models save time and which decisions need human judgement.
What I can help with
Workflows with agents
We set up a repeatable workflow: gather context, plan, implement, test, review. We practise it with Codex CLI, GitHub Copilot, Claude or whichever tool your team uses.
Testing and code review
We check model assumptions, review diffs and run tests. We also discuss agent permissions and responsibility for the code.
Context, privacy and local models
We look at what data each tool receives, how to limit context and when to consider a local LLM.
Costs, subscriptions and working practices
We compare token usage and pricing models, and match the subscription to the team’s needs and constraints. We decide what to delegate to an agent, when to check its work and when to stop an unsuccessful approach.