Design a ready-to-deliver workshop about Anthropic and Claude for the audience and setting below. Keep claims about products, capabilities, pricing, policies, and integrations limited to [APPROVED MATERIALS]; mark anything that needs current official verification. Do not invent case studies, customer results, product features, or safety claims. Audience and starting level: [AUDIENCE] Workshop goal and duration: [GOAL AND DURATION] Use cases participants need: [USE CASES] Approved materials and constraints: [APPROVED MATERIALS] Available format and tools: [FORMAT AND TOOLS]
Create a facilitator pack with: 1. A timed agenda that totals exactly [GOAL AND DURATION], including an opening, one short explanation segment, a live demonstration, hands-on practice, debrief, and close. 2. A slide-by-slide outline: title, one teaching point, suggested visual or live-screen action, and speaker notes. Limit each slide to one main idea. 3. Two realistic prompt demonstrations drawn from [USE CASES]. For each, show a weak first prompt, an improved prompt, the information it needs, and how to check the output. Use placeholders rather than fabricated business data. 4. One participant exercise with instructions, a worksheet template, expected output, and a peer-review rubric covering context, constraints, output format, factual support, and safe handling of sensitive information. 5. A facilitator checklist for account access, privacy boundaries, accessibility, timing, and fallback if the live demo fails. 6. A one-page takeaway with a reusable prompt structure and three follow-up practice tasks. Before answering, verify that the agenda duration adds up and that every activity serves [GOAL AND DURATION]. Flag claims not supported by [APPROVED MATERIALS]. Ask up to three clarifying questions only if a required input is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [AUDIENCE] | Describe participant roles, number of people, prior Claude or AI experience, and any accessibility needs. | 24 operations managers with basic AI familiarity; virtual session; two participants use screen readers. |
| [GOAL AND DURATION] | State the practical outcome participants should achieve and the total session length. | In 75 minutes, participants can draft a structured operations brief and evaluate it for unsupported claims. |
| [USE CASES] | List the work tasks participants want to improve with Claude. | Summarizing internal meeting notes, drafting SOPs, and turning incident notes into a stakeholder update. |
| [APPROVED MATERIALS] | Paste approved links, product notes, policy guidance, and claims you are allowed to make. | Our internal AI-use policy permits non-confidential examples only. Use official Anthropic documentation links supplied by the facilitator; do not discuss pricing. |
| [FORMAT AND TOOLS] | State whether the session is virtual or in person and what slides, accounts, breakout rooms, and devices are available. | Zoom, shared Google Doc, breakout rooms of three, and a facilitator Claude account for demonstration. |
How to use
- Paste approved product information and internal AI policy before asking for the workshop pack.
- Check that the agenda really totals the available minutes and that participants have enough time to practice.
- Replace all fictional demonstration placeholders with sanitized examples relevant to the group.
- Follow up with: “Create the participant worksheet as plain text that I can paste into a shared document.”
Variations
Executive briefing
Use for a short leadership introduction rather than a training session.
Create a [DURATION]-minute executive briefing on Claude for [AUDIENCE]. The decision or awareness goal is [GOAL], and the only approved facts are [APPROVED MATERIALS]. Produce a six-slide maximum outline with speaker notes, one carefully scoped demo concept, benefits tied to [BUSINESS PRIORITIES], risks and governance questions, and a proposed next step. Do not make product, ROI, or security claims beyond the materials. Check that each slide earns its place and flag facts requiring official verification.
Hands-on lab
Use for a practice-heavy session with a shared workflow.
Build a hands-on Claude lab for [AUDIENCE] to complete [WORK TASK] in [DURATION]. Participants can use [TOOLS AND ACCESS] and must follow [DATA RULES]. Provide a five-minute setup, numbered lab instructions, a sanitized input template, an expected-output rubric, facilitator troubleshooting notes, and a ten-minute debrief. Include checkpoints for source support, confidential-data handling, and human approval. Do not assume product features not documented in [APPROVED MATERIALS].
Prompt clinic
Use when participants bring their own prompts to improve.
Design a [DURATION] prompt clinic for [AUDIENCE] using Claude. Their common work tasks are [TASKS], and privacy limits are [DATA RULES]. Create a timed flow where participants bring a flawed prompt, identify missing context and output constraints, revise it, test it with safe sample data, and peer-review the result. Include a one-page scoring rubric and facilitator language for correcting unsupported claims. Use only these approved product facts: [APPROVED MATERIALS].
Tips
- Build the live demo around a sanitized work artifact; a generic poem or trivia example does not transfer to participants’ jobs.
- Reserve time to inspect output quality, not only to write prompts; participants need to practice finding unsupported statements and missing constraints.
- Keep product claims separate from prompt-writing instruction so a change in a feature or policy does not invalidate the entire session.
- For a mixed-experience group, give beginners a completed input template and advanced participants an optional edge-case challenge.
FAQ
How long should a Claude workshop be?
Sixty to ninety minutes is usually enough for one workflow, a demonstration, practice, and debrief. Longer sessions should cover distinct tasks, not add more lecture.
Can I use customer or employee data in a live demo?
Use only data permitted by your organization’s policy and the relevant service settings. Sanitized examples are safer for group sessions.
What makes a prompt exercise useful?
It should produce a work artifact participants recognize, with a defined input, expected output, and a way to judge factual support.