How to Set Up Claude Cowork the Right Way (So It Actually Does Your Work While You Step Away)
Don't want to figure this out alone? I walk members through every step inside the community. Join the Skool → skool.com/raycfu
Most people treat Claude like a chatbot. Type a question, get an answer, repeat. That's not how you get real value out of it.
Cowork is different. It lives on your desktop. It reads and writes files on your actual computer. It creates Word docs, builds spreadsheets with working formulas, installs specialist plugins for your exact job. And when it doesn't have enough info to do something well, it asks you instead of guessing and giving you polished garbage.
This guide will take you 30 minutes. After that, you'll have an AI workflow that saves you 5+ hours a day on research, writing, analysis, and whatever else eats your time.
What You Need Before You Start
Go to claude.com/download and grab the desktop app (macOS or Windows). You need a paid plan. Pro is $20/month. Max starts at $100/month for heavier usage. Open the app. At the top you'll see a toggle. Switch to the Cowork tab. That's it. You're in.
Step 1: Create Your Context Files (Minutes 0-10)
This is the step most people skip. Don't skip it. This is what separates "generic AI output" from "this actually sounds like something I'd write."
Create a folder on your computer called "Claude Context." Inside it, make three files. Save them as .md (markdown) files. You can use any text editor.
File 1: about-me.md
Write down who you are, what you do, your role, what success looks like in your work. Include your industry, the kinds of projects you work on, and what your day to day actually looks like. If you manage people, say so. If you're a solo founder, say that. The more specific you are here, the better Claude gets at everything else.
Example of what to put in here:
- Your name and job title
- What company or industry you're in
- What kind of work you do daily (proposals, reports, research, code, emails, etc.)
- One or two examples of work you're proud of
- Who you work with (clients, team, stakeholders)
File 2: brand-voice.md
This is how you communicate. Write down phrases you actually use. What tone sounds right to you. What sounds wrong. Paste in a few examples of writing you've done that you liked. If you hate corporate speak, say that. If you're casual and direct, show it.
Example of what to put in here:
- 3-5 sentences you've written that represent your voice well
- Words or phrases you use a lot
- Words or phrases you hate (e.g. "leverage," "synergy," "circle back")
- Whether you prefer short and punchy or detailed and thorough
- How formal or informal you are
File 3: working-style.md
This tells Claude how to behave when working with you. Do you want it to ask questions before starting? Do you prefer short outputs or long ones? What file formats do you like?
Example of what to put in here:
- "Always ask clarifying questions before starting any task"
- "Keep outputs concise unless I ask for detail"
- "Save documents as .docx unless I say otherwise"
- "Don't use bullet points unless I specifically request them"
- "When in doubt, give me options instead of picking one"
Pro tip: If you hate typing all this out, just talk it out using voice-to-text and clean it up after. Getting the context into text is what matters. How you get it there doesn't.
Once these three files are in your Claude Context folder, select that folder in Cowork by clicking the folder icon. Claude can now read everything inside it.
Your first prompt after this:
"Read all the files in this folder completely. Then give me a summary of what you know about me, how I work, and what context you have access to."
If Claude nails it, you're set. If something's off, fix the files now before you start real work.
Step 2: Set Global Instructions (Minutes 10-15)
Global instructions are permanent rules that load at the start of every Cowork session. Without them, Claude starts every conversation completely blank. No memory of who you are. No idea how you like things done.
Go to Settings, then Cowork in the desktop app. Click "Edit" next to Global Instructions.
Paste in the highlights from your three context files. Keep it tight. Think of this as the cheat sheet version.
What to include:
- Your name, role, and what you do in one or two sentences
- Your communication preferences in one or two sentences
- Your output defaults (file format, length, style)
- Your working style preferences (questions first, short vs long, etc.)
- Anything you never want Claude to do
Example:
"I'm [name], a [role] at [company/industry]. I prefer concise, direct outputs with no filler. Always ask clarifying questions before starting any task. Save documents as .docx unless I specify otherwise. Never use corporate jargon. When I give you a vague task, ask me what good looks like before you start building."
Save it. Now every Cowork session starts with Claude already knowing this. You never have to re-explain yourself again.
Test it by opening a new session and asking:
"Before we start any work, tell me what you know about me, how I like to work, and any standing preferences you're aware of."
If it reflects your instructions back correctly, you're good.
Step 3: Install a Plugin (Minutes 15-20)
Without a plugin, Cowork is a smart generalist. It can do a lot, but it doesn't know your industry's specific language, workflows, or what a good output looks like for your role.
Plugins fix that. They're pre-built skill packs designed for specific job functions. They add slash commands and structured workflows that make Claude way more opinionated about what to produce.
To install one:
- Click the "+" button in the chat bar
- Click "Plugins" to browse the library
- Or go to claude.com/plugins to see everything available
- Pick the one that fits your role and click Install
- Type "/" in any Cowork chat to see what commands the plugin added
Here are the main ones:
Productivity: task management, calendars, daily workflows. Good starting point for anyone. Marketing: content drafting, campaign planning, brand voice alignment. Sales: account research, call prep, outreach sequences, battlecards. Finance: financial modeling, analysis, reporting. Data Analysis: SQL, dashboards, dataset exploration. Legal: contract review, research, drafting. Product Management: specs, roadmaps, user stories. Customer Support: ticket handling, response drafting.
Pick one. Install it. Try a slash command. See how much more structured and useful the output is compared to a generic prompt.
Example prompts after installing:
Productivity: "/productivity:start Let's review what I need to get done today and set up my task list."
Marketing: "/marketing:draft-content Write a LinkedIn post about [topic]. Match the voice from my brand-voice.md file. Target [audience]. Goal: [what I want people to do]."
Data: "/data:explore I've uploaded a CSV in this folder. Give me a summary of what's in it, flag any anomalies, and suggest three analyses worth running."
Step 4: The One Line That Changes Everything (Minutes 20-25)
This is the most important step. It's one sentence you add to the end of every prompt:
"DO NOT start working yet. First, ask me clarifying questions so we can define the approach together. Only begin once we've aligned."
Here's why this matters so much.
Every other AI tool guesses when you give it something unclear. It picks an interpretation, runs with it, and hands you a polished output that answers the wrong question. You asked for a client proposal and it gives you a generic one. You asked for a summary and it summarizes the wrong things.
When you tell Cowork to ask first, it stops and generates a form. Multiple choice questions. Specific options. A structured set of questions that helps you think through what you actually want.
This does two things. First, you get way better output because Claude actually understands what you need before it starts. Second, you stop wasting time on revisions because the first draft is usually close to right.
You don't need to write perfect prompts anymore. You just need to say what you want and let Cowork figure out what it needs to know.
Your default opener for almost any task:
"I want to [YOUR TASK] so that [WHAT GOOD LOOKS LIKE]. First, read all uploaded files completely before responding. DO NOT start executing yet. Ask me clarifying questions to refine the approach. Only begin work once we've aligned."
Try it once. You'll never go back to writing long, carefully engineered prompts again.
Step 5: Connect One Tool (Minutes 25-30)
Connectors let Claude pull live data from your existing tools. No copy-pasting. No screenshots. No exporting files just to drag them into a chat.
Go to Settings, then Connectors in the desktop app. Browse the directory (50+ integrations including Slack, Google Drive, Notion, Figma, Asana, and more). Click a connector and hit "Add." Authenticate with that tool. Done.
You only do this once. After that, Claude can access live data from that tool in every Cowork session. This is free on all plans.
Pick the one tool you use most and connect it first.
Example prompts after connecting:
Slack: "Search my Slack messages from the last 7 days and give me a summary of anything I need to follow up on. Organize by urgency."
Google Drive: "Find the most recent document about [project name] in my Drive. Read it and tell me the three most important things I need to know."
Notion: "Pull my open tasks from Notion and help me prioritize them for today."
Things to Know
Cowork has no memory between sessions. Every new conversation starts fresh. That's why your context files and global instructions matter so much. They're your workaround for this limitation. Keep refining them over time and Claude gets better at your specific work every week.
Tasks stop if you close the app. Put your computer to sleep instead if something is running. The session survives sleep but not a full app close.
Usage goes faster than regular chat. Multi-step tasks with file reading and document creation use more compute than a normal conversation. If you're on Pro and doing heavy daily use, keep an eye on Settings then Usage.
Desktop only for now. No mobile, no browser version, no syncing between devices. If you work across devices, put your context files in a cloud-synced folder like iCloud or Dropbox so at least your files stay consistent.
It doesn't generate images. Use Gemini Imagen or another tool for that. Cowork is for documents, spreadsheets, presentations, and research.
You just read the full playbook. Most people will close this tab and never implement it. The ones who do usually hit a wall around the technical setup and quit.
Inside the Skool, I walk you through the exact build step-by-step, troubleshoot your setup live in the community, and share the scripts and templates I use to actually land paying clients.
If you want the shortcut instead of the long way around: