Three-level AI Academy

A structured three-tier AI training curriculum with audience, format, and content defined per level:

• Level 1 Basic — all employees; monthly online; covers AI basics, the company chat platform, PII / data residency regulation protection.
• Level 2 AI in Sales — new sales hires (Retail / Online); monthly online; AI sales-development tools (data search, call-quality dashboards, real-time prompts).
• Level 3 Claude Code Builders — product managers, team leads, key specialists, ambassadors; quarterly in-person; Claude Code, agent infra, vibe-coding, MCP integrations.

Generic "AI training" fails at scale — after one Q2-2025 generic training, 40% tried but only 15% kept using. Tiered training works because each level matches the audience's actual job-to-be-done: basic users get safety + the company chat tool; sales gets sales tools; builders get the deep stack. A wealth-management firm running this at enterprise scale hit 87% IT adoption and +50% YoY developer productivity.

1. Define three audiences and three formats — don't try to put one curriculum at three depths.
2. Level 1: 60–90 min monthly online — onboarding-style.
3. Level 2: function-specific monthly online — bring your real deals/cases, leave with a working workflow.
4. Level 3: quarterly in-person intensive — Claude Code hands-on, agent infra, ship a real workflow before leaving.
5. Add a Builders track for self-identified champions (Claude Code + MCP + vibe-coding).
6. Track per-level completion + post-class adoption (did they use the tool the week after?).

• Don't mix levels in one room. Putting builders and basic users in the same workshop drops engagement for both — builders are bored, beginners drown.
• Twice-weekly burnout signal. Operators who try the 2x/week 2-hour cadence past month 2 consistently report exhaustion. Either downshift cadence or split the audience by skill tier (60/30/10 above).