The Opening Ritual
Every Session Starts Here
The question:
"What's the worst prompt you wrote this week, and what did you learn from it?"
Instructions for Facilitator
- Weeks 1–3: Preston goes first. Model that bad prompts are expected and instructive.
- Keep it tight: 30 seconds per person. The point is normalization through repetition, not deep analysis.
- If someone hasn't used Claude: Respond with "That's useful data too. What got in the way?" No judgment. Just curiosity.
- By Week 4: Ask a champion to go first instead. This signals the ownership shift.
Why this matters: Psychological safety is built through repeated modeling. Every leader in the room learns that struggle is signal, and the facilitator is the first one to signal it.
Session Structure by Phase
Weeks 1–3: Preston-Led Foundation
Total time: ~35 minutes
- 2 min: Opening ritual
- 10 min: Activity walkthrough or demonstration
- 15 min: Hands-on practice (everyone works in Claude simultaneously)
- 5 min: Share one discovery or question
- 3 min: Preview next week + remind about weekly check-in
Goal: Establish the ritual, build confidence, normalize learning by doing.
Weeks 4–6: Co-Facilitated Transition
Total time: ~35 minutes
- 2 min: Opening ritual (champion goes first)
- 5 min: "Show and Tell" — 2 participants present their week's experiment (2.5 min each)
- 10 min: Group problem-solving on a challenge someone is facing
- 10 min: New technique or advanced application (Preston or champion)
- 5 min: Module-building check-in — "Where is your unit's training module?"
- 3 min: Preview + check-in reminder
Goal: Shift ownership to champions. Move from "I learned" to "I taught."
Weeks 7–8: Champion-Led Sessions
Total time: ~35 minutes
- 2 min: Opening ritual (champion leads)
- 10 min: "Show and Tell" — 3 participants present experiments
- 15 min: Group problem-solving or workflow redesign review
- 5 min: Module peer review — one leader presents their unit module draft for group feedback
- 3 min: Capstone preparation reminder
Goal: Champions own the session fully. Preston observes and advises only.
The Failure Wall
A Shared Space for Learning Together
The Failure Wall is a shared space (Monday.com board or Teams channel) where participants post failed prompts and what they learned. This is where the cohort learns at scale.
Ground Rules
- Preston and champions post first to establish the norm. Failure-posting is not optional; it's part of the culture.
- Celebrate the most instructive failure each week during the opening ritual. Name it, applaud it, explain what the cohort learned.
- Frame failures as data: "This prompt didn't work because..." teaches the whole cohort how to diagnose their own mistakes.
- Never critique a failure post. Only respond with "Here's what I'd try" or "I had a similar experience." This is not a place to judge; it's a place to build shared knowledge.
Facilitator tip: If the Failure Wall goes silent, it's a sign that psychological safety hasn't landed yet. Go back to modeling. Post your own failure, speak about it in the opening ritual, and invite a shy participant to share theirs (with permission first).
Leadership Module Accountability
Department / Team Leaders: Build Your Team Environment For Your Team and the Organization to Use
This pilot produces 10 leaders who shape the AI environment for their department. Each leader is responsible for defining what goes into their unit's Claude workspace: which organizational knowledge to load, which skills and projects are worth implementing, and which practices are worth streamlining. This is the capstone.
Environment Design Structure (The Scaffold)
Every leader's environment design addresses five elements:
- One real-work scenario from the unit's actual workflows (not a generic example)
- Governance tier classification for the scenario's data (public, internal, sensitive, restricted)
- A worked prompt example using CTCC (Context-Task-Content-Constraints)
- An output evaluation exercise: What's good? What needs revision? Why?
- A "What You CAN Do" permission list for the role (what's allowed, what's not, and why)
Timeline
- Weeks 1–3: Use Claude yourself. Identify your unit's highest-value AI use case.
- Weeks 4–5: Design your unit's environment. Identify the knowledge, skills, and projects your team needs. Bring questions to sessions.
- Weeks 6–7: Peer review. Present your environment design to the cohort for feedback.
- Week 8: Finalize. Your team inherits the intelligence layer you designed during Q1 rollout.
Facilitator Accountability
Check in on module progress every session starting Week 3. The weekly check-in (in your tracking sheet) captures this. Use this as your conversation starter:
"Where are you in your module? What's the real-work scenario you're using? What question do you have that we can solve as a group right now?"
If someone is stuck, pair them with another leader who's further along. This is peer mentorship in action and deepens psychological safety.
The Five Principles of Psychological Safety in This Pilot
- Struggle is signal, and signal is valuable. When something doesn't work, the cohort learns. Every failure that's shared is a data point that moves the whole group forward.
- Leaders model first. If you haven't shared a failure, you haven't earned the right to expect others to. This applies to Preston and champions most of all.
- Curiosity over judgment. The response to "I haven't used Claude this week" is "What got in the way?" and nothing else. No criticism, no impatience. Just understanding.
- Progress over perfection. A leader who used Claude once this week and learned something is ahead of one who used it ten times on autopilot. Learning is the metric, not volume.
- This pilot invests in YOU. AI fluency is a career asset that increases your value as a leader, inside DC CAP and beyond. You're building a skill that will define leadership in your organization.