Reading Time: 13 minutes

Hey Prompt Lover,

You asked for this one more than anything else.

After the last few issues went out, one message kept showing up in my inbox more than any other.

"When is the Cowork guide coming?"

It's here. Today. All of it.

But before I get into it, I want to say something real quickly.

The response to this series has genuinely surprised me. People sharing it with their teams. People replying with questions. People telling me it changed how they think about AI at work. That's the whole reason I write this thing and I don't take it for granted.

So thank you. Seriously.

Now let's get into what you actually came here for.

61% of marketers say this is the biggest marketing shift in decades.

Get the data and trends shaping growth in 2026 with this groundbreaking state of marketing report.

Inside you’ll discover:

Download your 2026 state of marketing report today.

Get Your Report

Out of the box, it's mediocre.

You open it, point it at a folder, type a request, get back something that sounds like it was written by a stranger who doesn't know you, doesn't know your work, and doesn't know what good looks like for what you do.

Most people at that point think the tool is overhyped. They close it and go back to whatever they were doing before.

What they don't know is that they were thirty minutes away from something completely different.

Because Cowork properly set up is not the same tool. It's not even close. The gap between those two versions is a folder structure, three files, and a block of instructions you paste once.

That's it.

So that's what I'm walking you through today.

I think a lot of people still don't have a clear picture so let me be direct about this.

It's not a chat window with a different name.

It's Claude with direct access to your computer's files. You point it at a folder, you describe what done looks like, and you walk away.

Claude reads your documents, creates new ones, builds spreadsheets with real working formulas, organizes files, and saves everything directly to your disk.

No copy pasting from a browser. No downloading. No reformatting. Finished work in a folder.

Anthropic built it on the same architecture as Claude Code — their developer tool. Cowork is the version for the rest of us. Same engine, no terminal required.

If your workday involves documents, data, reports, research, presentations, or any kind of file work — this is for you.

Here's where most people go wrong before they even start.

They point Cowork at their entire Documents folder. Or their Desktop. Or their home directory.

Don't do that.

Cowork gets read and write access to whatever folder you select. If something goes sideways — and sometimes it does, because it's still a research preview — you want the damage contained.

Create one dedicated folder. Call it Cowork-HQ or whatever makes sense to you. Everything lives inside that one folder.

Inside it, you want four subfolders:

context — where you'll put the files that tell Claude who you are.

rules — where you'll put Claude's operating instructions.

projects — where active work lives, one subfolder per project.

outputs — where Claude delivers everything it creates.

That separation matters more than it sounds. When you set up your global instructions in a few minutes, you're going to tell Claude that context, rules, and projects are read-only.

It can never touch those. The only place it writes is outputs. That means your setup files are always safe and your deliverables always end up in one place.

Build that folder structure before you do anything else.

This is the part that actually changes everything.

Three files. And I need you to hear this clearly — do not write these yourself.

When people write their own context files, they write LinkedIn bios. Polished. Generic. Completely disconnected from how they actually work.

Instead, open a regular Claude chat — not Cowork, just the normal chat — and let Claude interview you. Here's how to build each one.

File 1: about-me.md

This file captures who you actually are as a professional.

Not your job title.

Not your resume.

Who you actually serve, what you're actually working on, what good work actually looks like in your world.

Tell Claude this:

Answer every question. Be specific.

The more specific you are, the better every session gets from that point forward.

Save what comes back as about-me.md in your context folder.

File 2: voice-and-style.md

This is the one that kills the generic AI output problem.

This file makes everything Claude writes sound like you.

Same process. Open Claude chat. Tell it:

When Claude compiles it, ask it to write the file as rules, not descriptions. "Never use the word utilize" is more useful than "prefers simple language." That distinction matters.

Save it as voice-and-style.md in your context folder.

File 3: working-rules.md

This tells Claude how to behave in every single session.

Same interview process. Ask Claude to interview you about whether it should ask clarifying questions before starting or just execute, what file formats you prefer, how it should name files, what it should always do before starting any task, what it should never do without asking you first, and how it should handle vague instructions.

When it compiles the file, tell it to write everything as direct instructions. Not suggestions. Rules. "Always do X" and "Never do Y."

Save it as working-rules.md in your rules folder.

These three files are your cold start solution.

Without them, every session begins from zero.

With them, Claude starts every session already knowing you, already knowing your voice, already knowing your standards.

That one afternoon of setup pays back on every session after it.

Once your files are ready, open Claude Desktop, go to Settings, find Cowork, and click Edit next to Global Instructions.

Paste this exactly:

Click save.

You just told Claude how to behave in every session going forward without ever having to say it again.

This one changes the whole dynamic and most people skip right past it.

Without it, when you give Claude a task with any ambiguity, it guesses. And guesses give you output that's close but not quite right. You then spend time correcting it, going back and forth, rerunning things.

That's the frustrating version of Cowork.

AskUserQuestion makes Claude pause and present you with structured clickable multiple-choice questions before it starts.

Who is the audience for this?

What format do you want?

How long should this be?

You click your answers. Claude gets the context it needs. The output comes back right the first time.

Your global instructions already trigger it automatically. But for any specific task where you want to make sure Claude asks before acting, add this line to the end of your request:

Now that you're set up, here's where it actually earns its keep.

File organization. Point it at a chaotic Downloads folder and tell it to sort by type, date, or content. Claude reads the actual contents of files — it'll recognize that a randomly named PNG is a Stripe invoice and categorize it correctly without being told. What would take you two hours takes it ten minutes.

Research and reports. Feed it source material and a structure. It produces synthesized documents that follow your formatting rules, not a generic template. Meeting transcripts into action items. Scattered notes into a polished report. Multiple articles into a comparison brief.

Spreadsheets that actually work. Cowork creates Excel files with real formulas. VLOOKUP, conditional formatting, multiple tabs, category totals. Not CSVs you have to fix yourself. Drop receipt screenshots in a folder, ask for an expense spreadsheet, and it reads the images, extracts the data, and builds the sheet.

Presentations from scratch. Give it a brief and reference files. You get a real .pptx with slides, not a text description of what slides could look like.

Batch processing. Rename hundreds of files. Convert formats. Extract data from PDFs. Any repetitive file task that would take you an afternoon can usually be described in two sentences and finished in minutes.

Once the basic setup is solid, these three features extend what Cowork can do.

Connectors link Claude to your external tools — Gmail, Google Calendar, Slack, Notion, Google Drive, and dozens more. Set them up in Customize > Connectors. Once connected, Claude can pull live data from these tools during any task without you doing anything manually. Ask it to summarize your emails from the last week, check tomorrow's calendar, or pull a file from Drive. You set it up once. It's available in every session after that.

Plugins bundle skills, connectors, and workflows into ready-made packages. There's a growing library covering sales, marketing, finance, legal, HR, and data analysis. Browse them in the Customize menu and install with one click. Start with one that matches your role and use it for a week before adding more.

Scheduled tasks are the feature that makes Cowork proactive instead of reactive. Type /schedule in any Cowork session or create one from the Scheduled section in the sidebar. Set a cadence — hourly, daily, weekly — and Claude runs the task automatically as long as your desktop is on and the app is open.

Real examples people are running right now: morning briefings that summarize Gmail and Calendar before you sit down. Weekly expense reports generated from a receipts folder. Friday file cleanup that organizes Downloads. Competitor monitoring that delivers a research brief every Monday. You write the prompt once. It runs on autopilot.

Cowork Projects give each project its own workspace with dedicated files, instructions, and memory that persists across sessions. This means Claude remembers what you worked on last week without you re-explaining it every time.

Create a Project from the left sidebar. This is different from your context files — those cover your general preferences. Project memory covers task-specific context for that particular piece of work.

Dispatch lets you assign tasks from your phone. Pair the Claude mobile app with your desktop and you can message Claude a task while you're away from your computer. Claude runs the full workflow on your desktop — files, connectors, plugins, everything — and delivers the result when it's done.

You're on a train. You text Claude to compile a report from three spreadsheets on your desktop. You get back to your desk and it's done. That's Dispatch.

I'm not going to pretend this thing is perfect because it's not.

It eats usage fast. One Cowork session can burn through what twenty-plus regular chats would use. On the Pro plan at $20 a month, you'll hit limits quickly if you use it heavily every day. The Max plan at $100 a month is more forgiving. Monitor your usage in Settings before you get cut off mid-task.

Scheduled tasks only run when your computer is awake. If your laptop is closed at 6am, your morning briefing won't run. That's the biggest practical limitation right now. Keep this in mind when deciding what to schedule.

No cloud sync. Everything is local to one machine. You can't start a task on your laptop and check it on another device. Dispatch from mobile helps bridge this but the files themselves stay on one computer.

It still makes mistakes. Cowork can misread files, take a strange approach to a task, or produce output that misses what you needed. Review deliverables before sending them to anyone. The plan review step — where Claude shows you what it intends to do before executing — is your safety net. Use it every time.

That's the full setup.

One folder. Three files Claude builds by interviewing you. Global instructions pasted once. AskUserQuestion baked in. From there, add connectors when you're ready, install one plugin that matches your role, and schedule one task you're currently doing manually every morning.

Give it a real task this week.

Not a test prompt. A real piece of work you actually need done. See what comes back. That's the only way to know what this thing can do for your specific workflow.

The Claude Code guide is next for the developers and the people who want to go deeper.

After that, the phone to laptop setup — how to pair Dispatch properly, how to message your desktop from anywhere, and how to come back to finished work.

If you have questions about anything in today's guide, reply and ask. And if there's something specific you want covered in the Claude Code issue, tell me now.

See you in the next one.

— Prompt Guy

P.S. The Musk vs. OpenAI trial starts April 27th. That coverage is coming too. A lot happening right now and none of it is quiet.