Pillar 1: Master the Craft¶
v0.8 — Working Draft
This page is under active development. Content is directionally accurate but subject to revision. Suggest an edit →
"Build capability through doing, not curriculum."
The Argument¶
Most organizations respond to AI adoption by commissioning training programs. They build curricula, schedule workshops, certify cohorts, and track completion rates. This approach is grounded in an assumption inherited from industrial-era workforce development: that capability is a function of knowledge transfer, and knowledge transfer is a function of instruction.
That assumption is wrong — or at least radically incomplete.
Craft mastery in AI adoption follows the logic of skill acquisition research (Dreyfus & Dreyfus, 1986): competence develops through stages from novice to expert, and the transition from each stage to the next depends not on absorbing more content but on accumulating situated experience. A curriculum can move someone from novice to advanced beginner. Moving from advanced beginner to competent — and from competent to proficient — requires deliberate practice in real work contexts with real stakes.
The distinction matters because AI tools are not static knowledge domains. They are interactive systems whose effective use depends on prompt engineering intuition, output evaluation judgment, and workflow integration creativity — none of which can be adequately taught in a classroom. The gap between "completed the course" and "uses AI effectively in daily work" is not a knowledge gap. It is a practice gap.
Furthermore, the tools themselves change faster than any curriculum can track. By the time an organization has designed, piloted, and rolled out a training program on GPT-4's capabilities, the frontier has moved. Craft mastery requires building learning agility — the meta-capability to rapidly acquire proficiency with new tools as they emerge — rather than point-in-time knowledge of current tools.
This pillar draws on the apprenticeship model: learning happens through doing, coaching, and progressive challenge. Organizations that master AI adoption will be those that create environments where employees build capability through daily experimentation, peer learning, and structured reflection — not through LMS completion certificates.
In Practice¶
A professional services firm replaced its planned 40-hour AI training curriculum with a "100 prompts" challenge: every knowledge worker committed to using AI tools for 100 distinct real work tasks over 90 days, logging results and sharing learnings in weekly peer circles. Completion rates exceeded the prior training program, and — critically — post-challenge tool usage was 4x higher than post-training usage in a control group.
A healthcare system embedded AI "craft coaches" (experienced users, not IT trainers) in each department. These coaches did not deliver instruction; they sat alongside clinicians and administrators, co-working on real tasks. The approach generated both capability and trust simultaneously — clinicians could see AI applied to their actual workflows rather than generic demonstration cases.
A manufacturing company adopted "AI Fridays" — protected time for teams to experiment with AI tools on self-selected problems. Management provided no curriculum, only access, psychological safety, and a lightweight structure for sharing outcomes. Within six months, teams had independently discovered and validated over 30 high-value use cases that the central AI team had not identified.
The 4×1 Matrix¶
| Dimension | Example |
|---|---|
| Tools | Prompt libraries, shared workspace for AI experimentation, peer learning platforms |
| Processes | Structured "100 prompts" challenges, weekly craft circles, apprenticeship pairings |
| Behaviors | Daily AI tool use on real tasks, sharing failures and workarounds, coaching peers |
| Change Skills | Facilitating peer learning, designing practice-based challenges, coaching without instructing |
Diagnostic Questions¶
- Practice ratio: What percentage of your AI capability-building investment goes to classroom/e-learning versus structured on-the-job practice?
- Usage persistence: Do employees continue using AI tools at the same rate 90 days after training as they did during training? If not, you have a practice gap, not a knowledge gap.
- Peer learning infrastructure: Do employees have regular, structured opportunities to learn AI craft from each other — not from trainers or IT?
- Learning agility: When a new AI capability is released, how quickly does your workforce begin experimenting with it? Is this measured?