A guided setup for business owners and non-developers. By the end you'll have a Claude that knows who you are, what you're working on, and how you like to work.
What is this?
If you're typing the same context into every conversation (who you are, what you do, what project you're on, what tone you want), you're doing it the hard way. Claude Code lets Claude remember.
This guide walks you through setting up your own Claude workspace: a folder on your computer where Claude reads about you, your projects, and your preferences before every conversation. Think of it as onboarding a new assistant who never forgets a meeting.
One app, three modes. We'll use the most powerful one.
claude.com/downloadWhen you open Claude, you'll see three tabs at the top. Each is a different way of working. Knowing the difference matters. We'll go deeper on this in Episode 2.
Quick conversations, brainstorms, drafts. No file access. Fast.
Background agent. You give it a task, walk away, come back to a result.
Folder access, persistent memory, your CLAUDE.md. The mode that knows you.
One folder on your computer is going to become Claude's "home."
A few good options. Pick one:
Open Finder, go to your home folder, create a new folder called Workspace.
/Users/yourname/Workspace
Open File Explorer, go to your Documents, create a new folder called Workspace.
C:\Users\yourname\Documents\Workspace
Point the app at the folder you just made.
Workspace and click OpenWhen the folder loads, you'll see its name in a pill above the input bar. Any files Claude reads or writes from now on live in that folder.
This is the magic part. Claude is going to interview you and set up your workspace based on your answers.
Copy the text below and paste it into the Claude Code chat window:
I'm new to Claude Code and I'd like you to help me set up my workspace. Please walk me through a friendly interview — one question at a time — about who I am, what I'm working on, and how I like to work. At the end, create my CLAUDE.md, .memory/ folder, and project files based on what I tell you. Use the onboarding-bootstrap pattern: ask one question at a time, adapt to my answers, and write everything to disk when you're done.
Hit Enter. Claude will start asking you questions.
Claude will ask you 8–12 short questions, one at a time. Roughly:
When the interview is done, Claude creates a structure that looks like this in your Workspace folder:
You don't need to understand every file. Just know: Claude reads them on every conversation, so it always knows who you are.
Now bring in the work you've already got: drafts, briefs, notes, exports.
Open your Workspace folder in Finder/Explorer alongside Claude Code. Drag files or folders in.
Workspace/project-a/, Workspace/personal/, Workspace/clients/, whatever matches how you actually think about your work. Easier for you, easier for Claude.
Open the page → ⋯ menu → Export → Markdown & CSV → unzip → drag the folder in.
File → Download → Markdown (.md) or Plain text (.txt). Save to your Workspace folder.
Already markdown. Just copy your vault into a sub-folder of Workspace.
Open the file → Save As → choose .txt or copy-paste contents into a new .md file in Workspace.
Prove it works before you trust it with real work.
Claude should respond with specifics: your name, your role, your project, your preferences. If it gives generic answers, something didn't save. Open .memory/identity.md in any text editor and check.
From now on
Three habits that make this whole system work.
Don't keep one conversation going for weeks. Start a new chat for each new task. Claude re-reads your memory each time, so it'll catch up fast.
Before you close the chat, ask: "Update my memory with anything important from this session." Claude will write to .memory/ automatically.
If Claude gets your tone wrong, your facts wrong, or your style wrong, say so. "That's not how I'd say it. Update principles.md." The system gets sharper every week.
Open the folder, skim what's been written. Delete anything stale. You're the editor of your own assistant. Keep it tight.
An assistant with persistent memory of who you are, what you're working on, and how you communicate. Most people will never set this up. You just did. Three weeks from now, you'll wonder how you ever worked without it.
If something goes sideways
Open CLAUDE.md and .memory/identity.md in any text editor. If they're empty or missing, the bootstrap interview didn't finish. Just paste the bootstrap prompt again.
Open .claude/settings.local.json. The starter file pre-approves common safe commands. If you want to approve more, ask Claude: "Add permission for X to my settings."
In Claude Code, look at the title bar. It shows the folder path. On Mac you can ⌘+Click the path to open it in Finder. On Windows, copy it and paste into File Explorer.
Just delete CLAUDE.md and the .memory/ folder. Run the bootstrap prompt again. Nothing else gets deleted.
Ask Claude. "I'm trying to do X but I'm stuck because Y." It's better at troubleshooting your specific setup than any guide ever will be.
New practical guides for European SMEs adopting AI, delivered when they drop. No spam, unsubscribe anytime.
Want this done for you, not just explained? Vectimo runs the AI adoption end to end.
Work with Vectimo →