Why Claude feels dumb sometimes
Quick, honest take: Claude's default memory is a goldfish. Every new chat, it's forgotten who you are, what you do, your tone of voice, what projects you're working on, what you decided last time. So you spend the first 10 minutes of every chat re-explaining yourself.
It doesn't have to be that way. There are three layers of memory you can stack on top of Claude — none of them require code, all of them are free, and they get progressively more powerful. Layer 1 takes five minutes. Layer 3 is a power-user move. Most people don't even know layer 1 exists.
The whole thing in plain English
By the end of this guide, your Claude will know who you are, what you're working on, and how you talk — without you typing a single thing into the next chat. Layer 1 is the freebie everyone should do today. Layer 2 is for when you keep pasting the same context every chat. Layer 3 is for when you want a real second brain.
Layer 1: Built-in memory (5 minutes)
Claude has had a "memory" feature for a while now and most people have never opened it. It's a little settings page where you can write down everything Claude should know about you — your job, your business, your tone, your kids, your tools — and it carries that across every chat from now on. Five minutes, zero code, biggest payoff for the smallest effort on this page.
Where to find it
- Open Claude.ai. Click your avatar (bottom left).
- Settings → Personalisation (or "Memory" depending on your version).
- You'll see a big text field. That's your memory.
What to put in it
Treat it like the brief you'd give a brilliant new assistant on day one. Hit these:
The basics
Name, what you do, who your customers are, where you live (timezone matters). Two sentences max.
Voice rules
"Australian English. Direct, warm, occasionally cheeky. No corporate-speak. No 'dive into'. Lowercase first sentence is fine."
Active projects
"Building a membership called Wright Mode. Running a Techella event. Writing a podcast." 1-2 lines each. Update this every quarter.
Default rules
"Skip the preamble. Don't ask permission to start. Push back if I'm wrong. No hype emojis." This is the one that changes everything.
Bonus: import your ChatGPT history
If you used ChatGPT before Claude, you can export all your old chats from ChatGPT settings, then drop the file into Claude's memory. Claude reads through it and pulls out the patterns — your tone, your common projects, your interests. Way faster than writing it from scratch.
Layer 2: Context files (30 minutes)
Layer 1 is "everything Claude should know about you generally". Layer 2 is "everything Claude should know for this specific kind of work". This is where you stop pasting the same brand voice doc into every chat.
The move: for each kind of work you do regularly, write a one-page context file. Save it on your computer. Drop it into Claude whenever you start that kind of work. Or — even better — load it into a Claude Project and you never have to drop it in again.
The 5 context files most people need
Voice + tone doc
How you write. Words you love. Words you ban. Sentence structures that feel like you. 1 page. The single highest-leverage file you can build.
ICP / customer doc
Who you sell to. What they're stuck on. The exact words they use to describe their problem. What they've already tried. Where they hang out.
Product / offer doc
What you sell. What's in it. What's not. The price. The before/after transformation. Common objections.
SOP / process docs
One per repeated task. "How I do podcast intros". "How I write Instagram captions". "How I do client onboarding". Save them. Hand them to Claude.
Where to put them
Two options. Easy mode: a folder on your desktop called "Claude context" — drop the file in when you need it. Power mode: create a Claude Project for each kind of work, upload the relevant files into the project's knowledge base. Now every chat inside that project has them automatically. No more pasting.
Time to set up: about 30 minutes for the brand voice doc, less for the rest. Build the brand voice one first — it's the one you'll use every day.
Layer 3: Obsidian (the second-brain layer)
This one's a bigger commitment, and it's not for everyone. If you already use Obsidian, skip the intro — you know where this is going. If you've never heard of it, here's the short version: Obsidian is a free notes app that stores everything as plain text files on your computer. Which means Claude can read all of it.
Connect Claude to your Obsidian vault and suddenly Claude has access to everything you've ever written down. Every meeting note, every brain dump, every blog draft, every transcript, every book quote. It pulls from this when you ask questions. It gets smarter the longer you use it — because your second brain keeps growing.
Honest reality check
Layer 3 is brilliant but it's a system. You have to actually take notes in Obsidian for Claude to read them. If you're not already a notes person, this layer is a future-you problem — focus on layers 1 and 2 first. Most people get 80% of the value from layer 1 alone.
If you want to go for it
- Install Obsidian (free, Mac/Win/Linux). Make a vault — that's just a folder where your notes live.
- Connect Claude Desktop to your Obsidian vault folder via the desktop app's filesystem access.
- Tell Claude where the vault is in Layer 1 memory: "My Obsidian vault is at ~/Documents/MyBrain. Read from it whenever I ask about my notes, my past work, or my ideas."
- Start taking notes. Even rubbish ones. The system gets better the more it has to draw from.
What to do this week
If you're going to do one thing on this page, do this:
The 5-minute version
Open Claude. Settings → Personalisation. Paste in your name, what you do, your active projects, and 3 voice rules. That's it. That's the 80% win. Layer 2 is the next month's project. Layer 3 is for when you're ready.
Once you've done layer 1, every chat will feel different. Less re-explaining, more momentum. You'll wonder why you didn't do it earlier.
The one rule that matters
Update layer 1 every time something changes. Launching a new project? Add it. Pivoting? Update it. Quit something? Delete it. Stale memory is worse than no memory — Claude will keep helping you with stuff you stopped doing six months ago.
Want me to walk you through your own setup?
Inside the Wright Mode Membership I share my actual layer 1 memory file, the brand voice template I use for layer 2, and the Obsidian vault structure for layer 3. Steal mine. Adapt it. Skip a year of trial and error.
Wright Mode Membership
Weekly trainings, a vault of prompts and skills, and a room of women building real AI workflows together.
Claude Code Masterclass
The complete walkthrough for non-coders. Install, set up, and start building with Claude Code — from zero.
Follow me on IG
@wright_mode — that's where I drop the new prompts, skills, and workflows the moment I find them.