When the tool stops being personal and becomes how the team works
The first phase of AI adoption is personal: individuals prompt their tools, develop skills (both the LLM’s and their own), and consequently get faster. The transformation is the second phase – when it stops being personal, and review, integration, planning, and the way work flows through the team all shift because AI is in every loop. That second shift is the harder one, and it is where most of the value (and most of the friction) actually lives. And it is a human problem with an AI flavour, not a tooling one – coordination, hand-offs, agreeing on what “done” means when half the diff is generated. That is collaboration territory, exactly where our years of agile-coaching work come to bear.
This category is the applied, on-your-own-stack track: the three-day workshops where a team brings its real codebase and its real process and rebuilds them, rather than learning a tool in the abstract. Claude Code, Cursor, and GitHub Copilot in Applied format put the change where it has to stick – in the daily workflow, not the sandbox.
The individual speed-up is phase one, and the easy one; making it hold across a whole team – review, hand-offs, what “done” means when half the diff is generated – is the transformation. This is the engineering-floor counterpart to AI Strategy: not the boardroom portfolio, the daily process underneath it. The AI-Assisted Software Development path is the structured route through it.