Focus areas:
- Process documentation
- Vendor evaluation
- SOP drafting
- Data reconciliation
- Meeting summaries
- Cross-team briefings
Champion responsibilities
For each deep-dive, the department AI Champion:
Books the session 3 weeks out
Collects demo material (recent wins, current workflows, backlog items)
Runs the session
Documents outcomes: workflows to build, workflows to promote, workflows to retire
Reports to Steering Committee within 1 week
Outcomes tracked
For each deep-dive:
- Attendance %
- Number of new workflows identified
- Number of workflows promoted to production within 30 days
- Champion confidence score (survey)
- Team confidence score (survey)
Every new employee completes this within their first 30 days. Structured as five 45-minute modules.
Module 1 — What AI can and can't do here (Day 1)
Objectives:
- Understand what tools we've approved
- Know what the tools do well and what they don't
- Know our data classification rules
Content:
- Tour of the approved AI tools (chat, IDE, meeting assistant, department copilot)
- The 4-tier data classification: Public / Internal / Confidential / Restricted
- The tool tiers: which tools take which classes
- Live demo: routing an example task to the right tool
Assessment:
- 10-question quiz on data classification
- Must pass 90% to continue
Module 2 — Your first workflow (Week 1)
Objectives:
- Compose a good prompt
- Understand system prompts vs user prompts
- Use RAG (department copilot) effectively
Content:
- The 4-part prompt: Role, Task, Constraints, Format
- Live-write a prompt for a real task with the trainer
- How to use the department copilot: when to search KB vs when to reason
- How to iterate: what to do when the answer is wrong
Assessment:
- Complete 3 real tasks with AI assistance
- Submit prompts + outputs to trainer for review
Module 3 — Personal setup (Week 1)
Objectives:
- Configure your master system prompt
- Load the About Me block
- Set memory hygiene defaults
Content:
- Work through individual/03-system-prompts/ templates
- Configure ChatGPT Custom Instructions
- Configure Claude Project settings
- Configure Perplexity Spaces
- Configure Cursor / Copilot personal preferences
Assessment:
- Show your master system prompt to the trainer
- Show your Custom Instructions
- Demonstrate memory is set correctly
Module 4 — Guardrails and safety (Week 2)
Objectives:
- Know what NOT to paste into AI
- Know how to spot a hallucination
- Know how to report incidents
Content:
- Data classification examples — role-play "would you paste this?"
- Hallucination examples: fabricated citations, wrong facts, made-up features
- Verification patterns: cross-check, ask for sources, use grounded tools
- Incident reporting: what, when, how, to whom
Assessment:
- Complete 5 "spot the problem" case studies
- Discuss with trainer
Module 5 — Your department's playbook (Week 3)
Objectives:
- Deep dive on your department's copilot
- Learn your department's top 5 AI workflows
- Meet your AI Champion
Content:
- Facilitated by department AI Champion
- Walk through the department copilot's prompt library
- Practice with 3 real recent tasks
- Q&A
Assessment:
- Ship 1 real work output using AI in your first week post-training
- Peer or manager reviews the output
Continuing education
After the 5 modules:
- Weekly "AI office hours" (30 min, opt-in)
- Monthly "what shipped" showcase across the company
- Quarterly deep-dive on a specific topic
- Certifications for advanced users (see quarterly-refresh)
Every quarter, every employee attends a 60-minute refresh. This is how you keep pace with the field.
Why
New models drop. New capabilities emerge. Old habits go stale. Without a forcing function, employees stop improving.
Structure
Segment 1 — What's new in the field (15 min)
Big model releases this quarter
New capabilities: vision, voice, tool use, longer context
Notable industry shifts
Segment 2 — What's new in our stack (15 min)
Tools added, tools removed
Policy updates
New workflows launched
Retired workflows and why
Segment 3 — Team wins (15 min)
3-5 short spotlights from different departments
Real workflow, real impact, real numbers
Q&A with the builder
Segment 4 — What's next quarter (15 min)
Roadmap for the AI program
Upcoming trainings
New tools in evaluation
Requests for pilots
Owned by
Steering Committee (agenda + guest list)
AI Champions (spotlight preparation)
L&D (logistics)
Certifications
Employees who complete the quarterly refresh for 4 consecutive quarters earn an internal "AI Power User" cert. Bonus: paired with department Champion for advanced training.
Advanced tracks
Beyond the quarterly, three optional certification tracks:
Prompt Engineer track
6-week structured curriculum
Advanced prompting patterns
Eval design
Cost optimization
Agent Builder track
8-week hands-on
MCP server construction
Agent workflow design
Safety engineering
AI Product Manager track
10-week curriculum
Metrics, adoption, ROI
Roadmap and prioritization
Cross-functional coordination
Certifications are internally recognized and may be tied to promotion criteria per role.
Refresh recording
Every refresh is recorded. Employees who miss the live session must watch within 2 weeks.
Feedback loop
After every refresh:
- 5-question survey (usefulness, pace, requests for next time)
- Champion syncs to review feedback
- Adjustments feed next quarter's agenda