June 12, 2026 · 10:00–16:00 · Levallois-Perret · Paris
We are beyond pumped to carve out an entire day together to create the AI foundation for your business! There’s no better way to dissolve the overwhelm of AI than by building with an incredible group of entrepreneurs.
This is your Workshop Resource Center that has everything you need to prepare for the workshop, build all day, and continue to learn even after we wrap up together.
Before tomorrow, please double check that you have your Claude Pro (or Max) account setup. Make sure you’ve got everything you need to get started.
This session is taught live from slides. Download the deck below to follow along and reference after the day.
↓ Intro to AI Slides PDF · coming soon
Try it yourself
Paste this into Claude for a personalized tool comparison for your business:
Model Comparison Prompt
I'm a [describe your business/role in 1-2 sentences]. I'm trying to figure out which AI tools actually make sense for me to use. Can you help me understand the key differences between Claude, ChatGPT, and Gemini — focusing on the use cases and strengths most relevant to someone in my situation? I want a practical comparison, not marketing language.
10:50 – 11:10Claude Setup
▾
Get Claude Pro ready for business use in 20 minutes. Follow the steps below — by the end, Claude will know who you are, how you work, and what you expect from it.
1 · Download the Apps
Install Claude on your phone and desktop so it's always within reach. claude.ai/download
2 · Settings → Personal Preferences
Tell Claude how you want it to behave across every conversation. Open Settings → Personal Preferences and fill in the two fields:
Who you are — your name, role, business in 2–3 sentences
Your business — what you do, who you serve, what stage you're at
Tools you use — Notion, Canva, Zapier, whatever's in your stack
How you like to engage with Claude — direct, concise, push back when needed, no fluff
Your technical level — e.g. "non-technical but strategic; explain things like I'm a smart non-developer"
3 · Settings → Capabilities
Memory → turn ON both toggles — Claude remembers things across conversations so you don't have to repeat yourself. You can also import memory from other AI providers.
Artifacts → turn ON — Claude produces formatted documents, charts, and interactive outputs
File creation & code execution → turn ON — Claude can build and run things, not just write about them
4 · Settings → Privacy
Claude has opt-in features to share your data to improve its models — double-check these are OFF if you'd prefer not to contribute
Turn off metadata sharing / "Help improve Claude"
For a full list of the features and terms we covered, visit the Resources section.
11:15 – 11:30Prompting Deep Dive
▾
Four frameworks that make any prompt better — plus a meta-prompt for when you don't know how to start.
RTCF — Role, Task, Context, Format my favorite
RTCF Framework
Role: You are a [e.g. brand strategist, CFO, legal advisor].
Task: [What you need done — be specific].
Context: [Anything it needs to know — project files, website links, what's important or not].
Format: [What good output looks like — list, table, tone, length, etc.]
RTF — Role, Task, Format
RTF Framework
Role: You are a [specific expert/role relevant to your task].
Task: [Describe exactly what you need done — be specific].
Format: Please respond with [format: bullet list / numbered steps / table / short paragraph / etc.]
TAG — Task, Action, Goal
TAG Framework
Task: I need to [what you're working on].
Action: Please [specific action you want Claude to take].
Goal: The end goal is [what success looks like — who is this for, what should it accomplish].
DREAM — Define, Refine, Execute, Adjust, Maximize
DREAM Framework
Define: I need help with [task/problem].
Refine: More specifically, [add any constraints, audience, tone, or context].
Execute: Please [specific output you want — write, analyze, create, summarize, etc.].
Adjust: Avoid [anything you don't want — jargon, certain formats, specific approaches].
Maximize: The most important thing is [what matters most about the outcome].
ROTE — Role, Objective, Task, Expectations
ROTE Framework
Role: Act as a [relevant expert].
Objective: The goal is to [what you're trying to accomplish overall].
Task: Specifically, [what you need Claude to do right now].
Expectations: [Length, tone, format, anything you want more or less of].
Bonus: When You Don't Know Where to Start
Clarifying Questions Meta-Prompt
Before you start, ask me 3 clarifying questions that would help you give me a significantly better response on this task: [describe what you're trying to do in one sentence].
Tips that actually help
One prompt = one task — don't pack multiple jobs into a single message
Give context upfront — include everything it needs in the first message, not drip-fed across follow-ups
Tell it what good looks like — specify format, tone, length, and any constraints before it starts
Keep threads short — 10–20 responses max, then start a new thread; use handover files to carry context forward
Edit the original prompt when something's off — don't resubmit the same prompt and hope for a different result
Add "Ask me questions until you're 95% clear on the task" when building something complex
11:30 – 12:15Brand Project & Skill Files
▾
Follow these steps to build your brand skill file and set up your Claude Brand Project — so Claude always knows your voice, your audience, and your business.
Step 1 · Add the Skill Creator
This skill helps Claude build other skill files. Click Customize → Skills → "+" → Browse, then search for skill-creator.
Step 2 · Create a Brand Project
Click Projects → New and name it "Brand". This becomes your dedicated workspace for anything brand or content-related.
Step 3 · Gather Your Brand Assets
Copy the prompt below and paste it into a new chat inside your Brand Project — but don't hit enter yet. First, gather everything you want to feed Claude. The more you give it, the better the skill file.
Brand Skill Builder Prompt
Role: You are a brand strategist and copywriter who specializes in helping founders define their voice.
Task: Help me build a skill file for my brand — a document I'll upload to Claude so it can write in my voice across all content.
Context: I'll share my website, examples of my writing, brand assets, and any other materials I have ready. Ask me one category at a time and use my answers to build the file. Cover: my business and what I do, my ideal client, my communication style and tone, what to avoid, key messaging themes, and examples of language that does and doesn't sound like me.
Format: Compile everything into a structured skill file with clearly labeled sections. Make it specific enough that Claude could write a caption, email, or proposal in my voice without me having to explain my brand again.
What to pull together before you start
✦ Your website link
✦ Pricing documents
✦ Brand colors (hex codes), fonts, and logos
✦ Social media captions or posts you love
✦ Emails or newsletters
✦ Brand guides
✦ Pricing guides
Step 4 · Follow Claude's Lead
Once you hit send, Claude will guide you through a series of questions. Answer them and feed it the assets you gathered. The more specific you are, the better the output.
Step 5 · Audit Your Skill File
Open a new chat window within the Project and paste the prompt below with your skill file attached. Claude will flag anything vague, missing, or contradictory.
Skill Audit Prompt
Role: You are a brand strategist and editor who specializes in reviewing AI instruction files and conversion copy.
Task: I've just created a brand skill file. Find it and audit it.
Context: This skill file is the document I attach to Claude whenever I need branded content. I've built it based on my brand assets, writing examples, and answers to guided questions. I want it to be as sharp and specific as possible — not just for voice consistency, but for conversion.
Audit for:
- Messaging gaps (anything vague, contradictory, or missing entirely)
- Conversion blind spots (is it clear what I do, who it's for, and why they should hire me?)
- Audience clarity (does it name a specific person with specific problems, or is it generic?)
- Offer & pricing language (is it confident and specific, or hedged and unclear?)
- Voice consistency (does the tone hold up across all sections, or does it drift?)
- Anything that sounds like every other business in my industry
Format: For each issue found, tell me: (1) what the gap is, (2) why it matters, (3) a specific fix or question I should answer to resolve it. Then give me a revised version of the full skill file with your changes clearly marked. Be direct — I want to know what's actually missing, not what's working.
Step 6 · Make Adjustments
Incorporate Claude's suggestions into your skill file — or make a note to revisit it after the workshop. This is a living document; it gets better over time.
Step 7 · Upload Your Bonus Skill Files
Add these to your Brand Project or directly to your Skills folder (Customize → Skills). Download them below.
✦Copy-Editing Skill — use this to proofread anything Claude creates
✦AI-Isms Skill — filters out AI-sounding language so your content reads human
✦Sales Copy Skill — a layered copywriting skill for writing anything sales-related in your business
afternoon
Sessions
1:30 – 2:15Daily Morning Debrief Build
▾
By the end of this session, you'll have a scheduled automation that pulls from your actual tools and delivers a focused briefing every morning — before you open your inbox.
1 · Add your Connectors
Cowork → Customize → "+" → Browse Connectors. Add the tools you want Claude to pull from. Common choices: Google Calendar, Notion, Gmail, Slack.
2 · Set Connector permissions
Once added, click each Connector and adjust which tools Claude is allowed to use. Be deliberate — only grant what you actually need.
3 · Make a Project
Projects → New → name it "Morning Debrief" → add a short description of your role and what matters most in your day. This gives Claude the context it needs to filter signal from noise.
4 · Create a Scheduled task
Open a new chat in the Project → type /schedule → paste the prompt below. Claude will walk you through setting the time. Customize the instructions to your unique setup.
Morning Debrief Prompt
Every morning, pull from [YOUR DATA SOURCES — e.g. Google Calendar, Notion, Gmail, Slack] and deliver a focused briefing that includes:
- What's on my calendar today
- My top 3 priorities (pulled from [YOUR TASK SOURCE — e.g. Notion, a tasks file])
- Any urgent emails or messages that need a response
- One open question or intention to reflect on before I start work
Keep it under 300 words. No fluff. Write it like a smart assistant who knows I'm busy.
5 · Follow Claude's lead
Claude will ask for anything it can't pull automatically.
Tips for your Morning Debrief
Use this project to adjust the instructions as needed
For builds like this, there are always a few rounds of iteration to get it exactly where you want it
Adjust it seasonally to prioritize certain habits or projects
2:15 – 3:15Claude Cowork & Canva Automation
▾
Coming soonContent will appear here before June 12
3:15 – 3:45HD & Astro Bestie Project
▾
Coming soonContent will appear here before June 12
reference
Resources
What You Built Today
What
Where to Find It
How to Use It
Claude Business Setup
Claude.ai → Your profile settings
Automatically applies to all conversations
Brand Skill File
Your downloads folder → upload to any project
Attach when creating branded content
Copy-Editing Skill
Downloaded today
Attach when editing any written content
AI-Isms Skill
Downloaded today
Use when reviewing Claude output for human-sounding copy
Sales Copy Skill
Downloaded today
Attach for any sales or conversion-focused copy
Daily Morning Debrief
Claude → Scheduled Tasks
Runs automatically — check Claude every morning
Carousel Creator Project
Claude → Projects
Open and paste raw notes; outputs carousel copy
Astro Bestie Project
Claude → Projects
Ask anything about your chart or current transits
Terms
Claude-specific features and general AI terms — in plain language.
Claude Terms
▾
Chat
The standard Claude.ai interface. Great for Q&A, brainstorming, writing drafts, and one-off tasks. If you need one thing done, start here.
Cowork
A more hands-off mode where Claude can take actions on your behalf: scheduled tasks, longer workflows, file management. More powerful than Chat, with fewer guardrails. Use with intention and review what it does, especially when you're getting started.
Code
Claude's ability to build and run code. Zero experience required. Describe what you need in plain language and Claude handles the rest.
Skills
Reusable instruction sets. Think of them as recipes for Claude. Write one once and Claude knows exactly how to handle that task type every time. Example: a brand voice skill that makes all your copy sound like you.
Projects
Persistent context folders, similar to Custom GPTs. Add your instructions, files, and tone guidelines once. Claude carries them into every conversation in that project. Use one project per major workflow or client.
Connectors / MCPs
Model Context Protocol, originated by Anthropic and now an open standard. Lets Claude connect directly to your tools (Gmail, Notion, Calendar) without needing API keys or developer setup. Set permissions deliberately. I always have Delete blocked.
Claude Design
A visual design tool inside the browser version of Claude. It's capable, but it burns through tokens fast. On the Pro plan, one design session can eat a big chunk of your daily capacity. Save it for tasks where it's genuinely the right tool.
Scheduled Tasks
Automations that run on a set schedule without you triggering them. Example: a morning debrief that pulls your top priorities before you open your email.
Routines
Automated sequences where Claude takes a series of actions in order. Example: an end-of-week wrap-up that summarizes, drafts next week's focus, and files notes.
Artifacts
Formatted, downloadable outputs Claude produces: documents, charts, interactive tools, code files. Anything you want to save or share.
Live Artifacts
Artifacts that pull fresh data each time you open them. Think of them as mini dashboards: a pipeline tracker, a content calendar, a weekly metrics page.
Chrome Extension
Brings Claude into your browser so you can research, summarize, and draft without switching apps. Powerful, but treat it like giving someone access to everything you're browsing. Limit it to what it actually needs. Use cautiously.
Claude Models
Opus is the deep thinker. Best for strategic planning and complex analysis. It burns tokens at 3 to 5 times the rate of other models, so use it when depth really matters.
Sonnet is the creative workhorse and your best default for almost everything.
Haiku is faster and lighter. Good for quick lookups, simple reformatting, and high-volume tasks where speed matters more than depth.
Extended Thinking
Claude works through a problem step by step before answering, like showing its work. Slower, but more accurate for complex decisions or anything where getting it right matters more than getting it fast.
Effort
Controls how much processing Claude applies before responding. Higher effort means more thorough but slower. You'll see this as a slider or option when using extended thinking or longer tasks.
General AI Terms
▾
LLM
Large Language Model. The technology behind Claude, ChatGPT, and Gemini, trained on massive amounts of text to understand and generate language.
AGI
Artificial General Intelligence. Theoretical AI that could match or exceed human intelligence across any task. Not here yet, but this is what most AI labs say they're building toward.
Prompt
The instruction or question you give an AI. Better context gets you better results. That's really all there is to it.
Context Window
How much text the model can hold in mind at once: your conversation, uploaded files, instructions, all of it. Start fresh conversations for new topics so you're not filling the window with irrelevant history.
Token
The unit AI models use to process text, roughly three-quarters of a word. Context windows, pricing, and usage limits are all measured in tokens.
API
Application Programming Interface. The technical bridge that lets developers connect AI models to other software. If someone says they're "using the API," they're building something on top of the model.
Hallucination
When AI confidently states something false. It's pattern-matching without knowing what it doesn't know, so it fills in gaps with whatever sounds right. Always verify facts that matter.
System Prompt
Hidden instructions that shape how an AI behaves before you say a word. Every chatbot, assistant, and AI product has one. Project instructions and skill files are essentially your personal system prompt.
Agent / Agentic
AI that takes sequences of actions: searching, clicking, writing, sending, not just answering questions. Cowork mode is agentic. Treat it accordingly.
RAG
Retrieval-Augmented Generation. AI that searches a document library before answering, so responses come from your specific files rather than general training data. Useful for knowledge bases, SOPs, or client libraries.
Fine-tuning
Training a model on specific data to specialize it for a domain. Different from prompting. It changes the model itself, not just how you talk to it. Usually only relevant if you're building a product.
Automation
Using AI (or any tool) to handle a task without manual effort each time. Scheduled tasks, Zapier workflows, and recurring routines all count. Do the thinking once, then let it run.
Markdown
A lightweight formatting language Claude uses constantly. Asterisks make bold, hashes make headings, dashes make bullet points. You don't need to learn it, but knowing it exists explains why Claude's output sometimes looks like it has symbols in it.
Local AI
AI that runs entirely on your device rather than in the cloud. Nothing leaves your computer. Useful for sensitive data, though the models are generally smaller and less capable than cloud-based ones like Claude.
Prompt Frameworks
▾
RTCF Framework (Cheyanne's fav)
Role — Tell it what role to play (e.g. brand strategist, CFO, legal advisor) Task — What it needs to do Context — Anything it needs to know about what it needs to do Format — How do you want the output? What does good look like?
RTF Framework
Role — Tell it what role to play Task — What it needs to do Format — How you want the output (bullets, email, spreadsheet)
Chain of Thought
A prompting technique where you ask Claude to "think step by step" before answering. It dramatically improves accuracy on complex tasks — Claude reasons through the problem rather than jumping to a conclusion.