Français
Deutsch
The silent tsunami in IT teams
When GitHub Copilot reached one million paying users in January 2023, many IT project managers shrugged: "It's a tool for developers, it doesn't affect me." Two years later, that position is untenable. Generative code is redefining relationships within IT teams in profound ways, and PMs who don't understand these dynamics lose their ability to lead effectively.
How generative code changes team dynamics
Velocity explodes — and estimates become obsolete
A feature that took a senior developer 3 days can now be prototyped in 4 hours with Claude or Cursor. Your Sprint plans, historical velocity metrics, story point estimates — all must be recalibrated. I've observed 40–70% acceleration on certain types of development.
The "code vs no-code" boundary blurs
With generative code, a BA can now generate a data extraction script, create a test API, or automate a repetitive process — without being a developer. This role permeability creates new frictions and new opportunities.
Technical debt accumulates faster
AI-generated code is often functional but not always maintainable. Without rigorous code review, technical debt accumulates at the speed code is generated. The PM must budget refactoring time even in delivery sprints.
New skills for the IT Project Manager in 2025
| Skill | Why it's necessary now |
|---|---|
| Reading AI-generated code | To validate quality, not rewrite |
| Understanding tools (Copilot, Cursor, Claude Code) | To have informed conversations with the team |
| Calibrating post-AI estimates | Historical velocities no longer apply |
| AI code quality governance | Defining review standards |
PM positioning as value architect
If generative code accelerates the "what to build", it doesn't answer "what to build first" or "why build this rather than that". This is precisely where the project manager and Product Owner see their value increase: strategic prioritisation, trade-off management, business objective alignment. These are not skills that AI can replace.