Working with AI
AI assistants are now a standard part of day-to-day development work. At its core, this is about using them well: giving good context, being explicit about role/task, constraints, and examples, and refining a prompt iteratively instead of accepting the first answer. Just as important is critically judging what comes back — catching hallucinations, outdated knowledge, or insecure suggestions — and knowing what you should never hand over to an AI tool, such as secrets or confidential data.
Once you're comfortable with that, you can go further: chaining AI into an agentic workflow (analyse → propose → refine → verify) instead of firing off isolated prompts, wiring it into your IDE, CLI, or CI so it's a genuine part of your development loop, and documenting when and how you used it so the work stays auditable.
Starting Points
- Pereira, S. (2025). Generative AI for Software Development. O'Reilly.
- LeMaire, C., & Abshire, B. (2025). AI for Everyday IT: Accelerate Workplace Productivity. Manning.
- SWEBOK Guide — "Software Engineering Professional Practice" and "Software Engineering Tools and Methods" chapters.
- Official documentation and responsible-use guidelines for the AI tools you use (e.g. GitHub Copilot, Claude, ChatGPT).
Key Points
- You use AI tools deliberately for common tasks (generating or explaining code, drafting documentation, suggesting tests, writing scripts or queries).
- You write clear, structured prompts with context, constraints, and examples, and refine them iteratively.
- You critically evaluate AI output for correctness, security, and plausibility, and recognise obvious hallucinations.
- You describe appropriate and inappropriate uses of AI in everyday development work.
- You apply responsible-use guidelines: you never share secrets or confidential data with an AI tool, and you document when and how AI was used.
- You design and run multi-step AI workflows for realistic tasks (adding a feature, refactoring, improving tests or documentation), rather than relying on single ad hoc prompts.
- You integrate AI tooling into your local development loop (IDE companions, CLI pipelines, lint hooks) so its use is observable, not hidden.
- You review AI-generated code, tests, and docs critically for correctness, security, and maintainability, and fix or reject them as needed.
- You reflect on how AI assistance changes your own workflow, and can propose concrete team guidelines for effective, safe use.