Which Claude is the right Claude?
When you sign up at claude.ai, you'll see a chat interface and assume that's it. It's not. Anthropic ships Claude in three different shapes, and picking the wrong one is the #1 reason beginners stall out. Here's the cheat sheet.
💬 Claude.ai (web)
What it is: The browser version. Open a tab, type, get an answer.
- Best for: Quick questions, writing, analysis
- Skill level: Day 1, no setup
- Use when: You're on someone else's laptop or just need a fast answer
🖥️ Claude Cowork (desktop)
What it is: A native Mac/Windows app that can see your screen, read your files, and run as a "coworker."
- Best for: Your actual day-to-day work
- Skill level: Beginner-friendly, more powerful
- Use when: You want Claude to actually do things, not just talk about them
⌨️ Claude Code (terminal)
What it is: A CLI tool that runs in your terminal and can edit files, run scripts, and build automations.
- Best for: Automations, building skills, file work
- Skill level: A bit techier — but worth it
- Use when: You're ready to graduate from chat to "Claude as an operator"
📱 Claude mobile app
What it is: The iOS/Android version of claude.ai. Same as the web, in your pocket.
- Best for: Voice notes, on-the-go thinking
- Skill level: Drop-in
- Use when: You're walking, driving, or in the school pick-up line
My honest take for week one
Start with Claude.ai in the browser AND install Cowork on your desktop. Use the web for quick thinking, use Cowork when you want it to read a file, summarise a transcript, or run something for you. Leave Claude Code until you're comfortable with the first two — it's powerful but the learning curve is real.
Free vs Pro vs Max — what to actually pay for
Anthropic's pricing page can feel like a wall. Here's the only thing that matters: how often do you hit the "you've reached your limit" message? That's the signal to upgrade — not the feature list.
Free ($0)
Great for testing the waters. You get the latest Claude model with limited messages per session. You'll hit the cap fast if you're using it for real work — usually after 10-15 messages. Use this to decide if Claude clicks for you.
Pro (~$20/mo)
The sweet spot for 95% of solo founders. 5x more usage than Free, Projects unlocked, longer conversations, larger file uploads, access to the desktop app. If you use Claude more than twice a week, get Pro.
Max (~$100–$200/mo)
For heavy users — content creators running multiple workflows, agencies, anyone using Claude Code daily. 5x–20x Pro's usage. Only upgrade when Pro starts feeling restrictive. Don't pay for capacity you won't use.
Team / Enterprise
For multi-person businesses. Shared Projects, admin controls, central billing. Skip unless you have 2+ people actually using it daily. Solo founders waste money here.
My honest take on upgrading
Start on Free for a week. If you keep hitting the limit before lunch, upgrade to Pro. Stay on Pro for at least 2 months before considering Max — most people don't actually need it. The best signal is when you genuinely think "this would be worth $200 for what it just did for me" — not when an upgrade prompt nudges you.
Your 10-minute account setup
Most people skip these steps and wonder why Claude feels generic. Do these once, and every conversation gets sharper from here on out.
Sign up at claude.ai
Use a real email you check daily — this is where confirmations and security stuff land. Skip the "sign in with Google" if you can; a direct email/password account is easier to manage long-term.
Set your custom instructions
Settings → Profile → "What should Claude know about you?" Add 3-5 lines: your name, what your business does, who your clients are, your tone preferences (e.g. "Australian English, direct, no fluff"). Claude reads this on every conversation.
Turn on memory
Settings → Memory → enable. From now on, Claude remembers context from past chats (your business, your style, your projects), so you stop re-explaining yourself every time. Already been using ChatGPT or another AI? You can bring its memory across too — that's all in Section 5.
Create your first Project
Top-left → Projects → New Project. Name it after a real workstream (e.g. "Content", "Client Onboarding", "Email Marketing"). Add a project description with the goal and context. Every chat inside this project inherits that context automatically.
Install Claude Cowork (desktop)
Go to claude.com/download and grab the Mac or Windows app. Sign in with the same account. Pin it to your dock — this is the version you'll actually use for work, not the browser tab.
Connect Google Drive (optional but powerful)
Settings → Connectors → Google Drive. Now Claude can read your Docs, Sheets, and Slides directly. Massive time saver when you're working with client notes, briefs, or content drafts that live in Drive.
⏱️ Don't go down a rabbit hole here
Set these six things and stop. You don't need to explore every menu, integration, or beta feature today. Better to actually use Claude for a week and notice what's missing than to pre-optimise settings you'll never touch.
The 5 settings that change everything
If you only do one section of this guide today, do this one. These are the settings that make Claude feel like your assistant instead of a generic chatbot.
1. Custom instructions (Profile)
What it does: Tells Claude who you are and how you like to communicate, on every chat.
Set it to:
- Your name + business + who you serve
- Tone (e.g. "warm but direct, Australian English, no corporate jargon")
- What you DON'T want (e.g. "no em-dashes, no 'dive in', no symmetrical three-word phrases")
- How much context to assume ("I'm an AI strategist — skip basic explanations")
2. Projects (with description + files)
What it does: Groups related chats under one persistent context. Add files once, every chat in the project can read them.
Set it up like this: One Project per workstream. Upload your brand voice doc, ICP description, or templates once. Now every chat in that project knows your voice without you re-explaining it. Saves hours a week.
3. Memory (with the right scope)
What it does: Lets Claude remember details across separate chats — your business, ongoing projects, preferences.
The hidden setting: Inside Memory, you can review and edit what Claude has stored. Do this monthly — it builds up junk like "user is currently testing X" that goes stale. Delete anything that's no longer true so Claude isn't acting on old context.
4. Artifacts (on by default — keep it on)
What it does: Opens a side panel for anything substantial — long documents, code, structured outputs — so you can iterate on it without cluttering the chat.
How to use it: When Claude writes you something long (a script, an email sequence, a strategy doc), it lands in the Artifacts panel. You can ask Claude to edit specific parts ("rewrite section 2 to be punchier") and it'll update the artifact in place. Stop copy-pasting between Claude and Google Docs.
5. File uploads (PDFs, images, transcripts)
What it does: Drag-and-drop any file into a chat — Claude reads it instantly. PDFs, screenshots, CSVs, Word docs, transcripts.
Underused trick: Screenshot anything confusing (a dashboard, an email, a competitor's landing page) and drop it in with "what is this telling me?" Faster than describing it in words. Works incredibly well for client work and competitor analysis.
Bring your memory across — and clean it up
If you've been using ChatGPT, Gemini, or another AI for a while, it knows things about you — your business, your tone, your projects. You don't have to start from scratch. Claude can import that history in one go, so it walks in already knowing who you are. Here's how to bring it over, plus how to stop your memory turning into a junk drawer.
First, what "memory" actually is
Memory is the stuff Claude remembers between chats — your name, what your business does, how you like things written, what you're working on. It's different from your chat history (the actual conversations). When you import, you're moving the memory and preferences, not every old conversation. That's the part that's worth carrying over anyway.
Importing from ChatGPT (or any other AI)
This takes about five minutes and works the same whether you're coming from ChatGPT, Gemini, or Copilot.
Ask your old AI to hand over what it knows
Open ChatGPT (or whatever you've been using) and paste this:
"I'm moving to another AI tool and need to export what you know about me. List everything: who I am, what my business does, my projects and goals, my communication style and preferences, and any context you've picked up from past chats. Put it all in one code block so I can copy it easily."
It'll spit out a tidy summary. Copy the whole thing.
Open Claude's import flow
In Claude: Settings → Capabilities → Memory → "Start import." (You can also click "Get started" on the import-memory card if it pops up on your home screen.) This works on Free, Pro, Max, and Team — on web and the desktop app.
Paste it in and hit "Add to memory"
Drop the copied text into the box and submit. Claude turns it into individual memory entries — give it up to 24 hours to fully process. That's it. Claude now knows your business without you re-explaining it.
⚠️ Read what you're importing first
Skim the export before you paste it. Old AIs collect a surprising amount of stale or personal-but-irrelevant detail ("user is planning a trip in March", "user was testing a recipe app"). Delete anything that isn't true anymore or isn't about your work — you want Claude starting clean, not inheriting someone else's clutter. This feature is still experimental, so it won't always grab everything perfectly — check the result in Section "cleaning up" below.
Cleaning up your memory (do this monthly)
Memory is brilliant if you keep it tidy. Left alone, it builds up old context that quietly skews everything Claude does for you. A five-minute monthly tidy keeps it sharp.
🧹 The 5-minute memory tidy
Quick shortcut: just ask Claude
You don't even have to dig through menus. In any chat you can say "What do you currently have stored in your memory about me?" to see it, or "Forget that I'm working on [old project]" to clear something out. Talking to it is often faster than the settings panel.
Back it up before you make big changes
Before a big clean-out (or if you ever want a copy), ask Claude: "Write out your memories of me verbatim, exactly as they appear in your memory." Save that text in a note or doc. Now you've got a backup you can re-import or refer to — handy if you tidy too aggressively and want something back.
How to actually prompt Claude
Beginner prompts are too vague. Pro prompts give Claude something to work with. The difference isn't being "good with AI" — it's giving the same context you'd give a human teammate on their first day.
❌ Generic prompt
"Write me an Instagram caption about productivity."
Claude gives you something. It's fine. It sounds like everyone else's caption. You scroll past it.
✅ Context-rich prompt
"I'm an AI strategist for women founders. Write an Instagram caption (under 220 words) about why most productivity advice fails women running businesses. My audience is overwhelmed solopreneurs juggling kids + clients. Tone: direct, warm, no corporate-speak. Hook with a contrarian opinion."
Claude has a brief. The output is usable.
🥪 The Context Sandwich (steal this framework)
Every prompt I write has three layers. Once you internalise this, you'll never write a bad prompt again.
1. Role + situation
Who you are, what you're working on, who it's for. "I'm a [X] writing for [Y] who [Z]."
2. The actual ask
What you want Claude to do. Be specific about format, length, tone. "Write 3 versions of [thing] in [format]."
3. Constraints + examples
What to include, what to avoid, what "good" looks like. "Don't use X. Match the tone of [example I'm pasting]."
4. (Bonus) Show your work
Ask Claude to think out loud before answering. "Before you write, list 3 angles and pick the strongest."
⚡ The 3 prompt upgrades that compound
- "Be specific about what you'd push back on." Makes Claude challenge your idea instead of just agreeing.
- "Show me 3 versions: safe, bold, and weird." One prompt, three angles to compare.
- "What context am I missing that would make this better?" Forces Claude to ask the smart questions you didn't think to answer.
Your first 5 use cases — pick one and start today
Don't try to "use Claude for everything" in week one. Pick the ONE use case that solves the most annoying part of your week, get a win, then expand. Here are the five I'd start with — each comes with a copy-paste starter prompt.
1. Inbox triage
Paste in your unread inbox subject lines + senders. Ask Claude to flag urgent, summarise newsletters, and draft replies for routine messages.
"Here's my inbox from the last 24h. Group into: urgent (reply today), context (read this week), and ignore (newsletters, promos). Draft 2-line replies for the urgent ones."
2. Voice notes → action items
Record a voice memo while walking. Drop the audio file (or transcript) into Claude. Get a clean to-do list and a draft of whatever you were thinking through.
"This is a voice note of me thinking about [topic]. Extract: 1) the actual decision I made, 2) action items with owners, 3) any open questions I should follow up on."
3. Transcript analysis
Drop in a Fathom or Zoom transcript. Ask for client objections, key quotes, follow-up actions, or content ideas — all from one call.
"Here's a client discovery call. Pull out: 1) their actual problem in their own words, 2) what they said they've tried before, 3) 3 content ideas this conversation sparked, 4) my next 3 follow-up actions."
4. Content repurposing
Take one long-form piece (a blog, YouTube transcript, podcast) and turn it into 5 social posts, an email, and 3 hooks for next week.
"From this YouTube transcript, give me: 3 IG carousel ideas, 2 LinkedIn posts, 1 email broadcast, and 5 hooks I could use as standalone Reels. Match my voice [paste sample]."
5. Research + first draft
Use Claude to scope a new offer, audit a competitor, or pull together a strategy doc from scratch. Saves the blank-page paralysis.
"I want to launch [offer] to [audience] in [timeframe]. Audit: 3 competitor offers (research from web if needed), 3 angles I could differentiate on, and a draft launch sequence. Be honest about gaps."
🎯 My pick for week one
What NOT to do (beginner traps)
After teaching hundreds of women founders to use Claude, these are the same mistakes I see in week one. Avoiding them puts you 6 months ahead of everyone else stumbling around.
Trap 1: Treating Claude like Google
Google answers "what's the capital of France." Claude can do that, but you're wasting it. Claude shines when you give it a problem with context — "I'm rewriting my homepage, here's the current draft, what's weak about it?" — not single-word lookups.
Trap 2: Dumping zero context, expecting magic
"Write me an email" gets you a generic email. "Write me a re-engagement email to my list of women solopreneurs who haven't opened anything in 60 days; tone is warm-but-direct; offer is a free 15-min audit" gets you something usable. The context isn't extra work — it IS the work.
Trap 3: Asking for "ideas" with no constraints
"Give me 20 content ideas" produces 20 mediocre ideas. "Give me 5 content ideas about pricing for women founders who undercharge — must be controversial, must work as a 30-second Reel" produces 5 good ones. Constraints force creativity.
Trap 4: Ignoring memory + projects
Every chat starts fresh unless you set this up. If you find yourself re-explaining your business or pasting the same voice doc into every conversation, you skipped Section 3. Go back. 10 minutes now saves you hours every month.
Trap 5: Copy-pasting outputs blindly
Claude is a brilliant first-draft machine, not a final-draft machine. Read everything before it goes out under your name. AI-tells (em-dashes, "dive in", symmetrical phrases, weirdly polished tone) are easy to spot — and they tank trust with your audience.
⚠️ The biggest trap: trying to use it for everything at once
The instinct in week one is to AI-ify your whole business. Don't. Pick ONE use case from Section 7, do it for 7 days, notice what works, then expand. Compounding habit beats heroic overhaul every time.
What to do instead — one small thing a day
You don't have to remember all of this. I've turned the whole guide into a simple 7-day plan below — one task a day, with the exact prompt to paste. Just scroll down and start with Day 1. 👇
Your 7-day plan — one thing a day
Everything above in one place, broken into days so you never have to decide what's next. Do one card a day. That's the whole instruction. Don't read ahead, don't try to do three days at once — just today's card. Each one takes 5–15 minutes and builds on the last.
How to use this: Pick a start day and go in order. If you miss a day, no guilt — just pick up where you left off the next time you sit down. The goal isn't speed, it's the habit. Where there's a prompt, copy it straight into Claude and swap in your own details.
Tell Claude who you are
Open Settings → Profile and fill in your custom instructions. This is the single highest-leverage thing you can do — it makes every future chat sound like you instead of a generic bot. (This is Section 3, step 2 + Section 4, setting 1.)
Then copy the result into Settings → Profile. Done for today.
Turn on memory + bring your history across
Switch on memory (Settings → Capabilities → Memory). If you've used ChatGPT or another AI, import what it knows about you so you don't start from scratch. (Full steps in Section 5.)
Copy the result → in Claude, Settings → Capabilities → Memory → Start import → paste. If you've never used another AI, just skip the import — memory's still on.
Make your first Project
Create one Project for a real part of your work — "Content", "Client Work", whatever you do most. Drop in your brand voice doc or a sample of your writing. Now every chat inside it already knows your context. (Section 3, step 4.)
If it gets anything wrong, fix your profile or add a file. That's the whole task.
Do ONE real task — your voice notes
Time to actually use it. Record a 2-minute voice memo of you thinking out loud about something on your plate, then let Claude turn it into a clean list. This builds the habit of talking to Claude. (Use case #2 in Section 7.)
That's it. One real win. Notice how much faster that was than typing.
Learn the Context Sandwich
Today's about better prompts. The trick is giving Claude context like you would a new teammate: who you are, the ask, and the rules. Try it on something you actually need this week. (Section 6.)
Same structure works for emails, captions, offers — everything. Keep this one.
Tidy your memory
A quick clean keeps Claude sharp. Open your memory, delete anything stale or finished, fix anything wrong. Five minutes now saves you weird, off-base answers later. (Section 5, the memory tidy.)
Then in Settings → Capabilities → "View and edit your memory", delete what's stale. Tell Claude "forget [old thing]" for anything quick.
Look back — and pick your next ONE thing
No new setup today. Just notice what worked and choose the single use case you'll lean on next week. The whole game is one habit at a time, not doing everything at once. (Section 7 + the big trap to avoid in Section 8.)
Pick that one thing. Use it daily next week. That's how Claude quietly becomes part of how you work.
If you only remember one rule
One small thing a day beats a big overwhelming overhaul every single time. You do not need to master Claude this week — you just need to keep showing up. Seven small wins and it's already part of your business.
Ready to go deeper?
When you're ready to level up from "I use Claude" to "Claude runs half my business" — start here.
Wright Mode Membership
Join a community of women entrepreneurs implementing AI and automation in their businesses. Live calls, templates, and ongoing support.
Claude Masterclass
Learn how to use Claude like a pro. From prompting fundamentals to building real workflows that save hours every week.
Claude Code Masterclass
Go beyond the chat interface. Build automations, process data, and create tools with Claude Code — no developer experience needed.