Course Overview
AI coding tools make teams faster. They also make teams fragile. When developers stop writing code from scratch, stop reading code critically, and stop questioning AI suggestions, the speed gains come at a cost that surfaces months later: architecture nobody understands, bugs nobody can diagnose without AI, and juniors who never learned the fundamentals.
This three-day course is for teams that want the productivity gains without the hidden costs. It teaches a disciplined approach to AI-assisted development grounded in extreme programming and test-driven development – practices designed for exactly this situation: high-velocity environments where engineering rigour is the difference between sustainable speed and accumulated risk.
You will learn to identify the antipatterns that plague early AI adopters – magic thinking, skills atrophy, quality degradation, “AI whisperer” concentration – and build concrete countermeasures. Through extended hands-on exercises, you will practise test-driven development with AI as your implementer, apply structured pair programming with defined human/AI roles, and design team practices that preserve craftsmanship alongside automation.
The course goes beyond individual technique into team-level concerns: metrics that distinguish real productivity from velocity theatre, risk indicators that signal degradation before it becomes visible in production, and adoption strategies with explicit guardrails. You leave with the ability to accelerate your team’s development without silently eroding the engineering culture that makes acceleration sustainable.
Learning Objectives
- Identify dangerous antipatterns in AI-assisted development: magic thinking, skills atrophy, quality degradation
- Apply test-driven development as a control mechanism for AI-generated code
- Structure human-AI pair programming with explicit role definitions and red-flag recognition
- Design team practices that preserve engineering fundamentals alongside AI adoption
- Establish metrics that balance velocity gains against quality and resilience risks
- Implement phased AI adoption with guardrails for critical code paths
- Build team resilience: skill preservation programmes, rotation practices, knowledge distribution
Topics Covered
- The Hidden Costs of AI Speed – Why early adopters lose engineering discipline, and the antipatterns that cause it: magic thinking, skills atrophy, Frankenstein architectures
- Test-Driven Development with AI – Red-green-refactor with AI as implementer; quality gates for AI-generated tests; testing antipatterns specific to AI collaboration
- Human-AI Pair Programming – Structured role models (navigator/driver, reviewer/implementer); when to reject AI suggestions; recognising deprecated patterns and hidden complexity
- Preserving Engineering Craftsmanship – AI-free practice sessions, debugging without AI, system design from first principles; preventing the erosion of skills your team cannot afford to lose
- Testing AI-Generated Code – Trust boundaries, property-based testing, critical validation points; the difference between “tests pass” and “code is correct”
- Metrics and Risk Monitoring – Velocity vs quality indicators, signals of degradation (declining review depth, growing tech debt, reduced architectural discussions), building visibility into AI adoption health
- Building Resilient Teams – Preventing “AI whisperer” concentration, career development in the AI era, rotating AI-free sprints, knowledge distribution practices
- Safe Adoption at Scale – Phased implementation, AI-free zones for sensitive code, mandatory human review for critical paths, incident analysis, continuous antipattern review
What You Get
- Three days of intensive hands-on practice with realistic coding scenarios
- Techniques for identifying AI adoption antipatterns in your own team
- Practical team practices you can implement immediately after the course
- A metrics framework for monitoring AI adoption health
- Certificate of completion