⏱️ Your First 30 Minutes With Claude AI

You don’t need five tools, a course, or a YouTube rabbit hole. You need one tool, set up properly, and half an hour with a cuppa. That’s this guide.

By minute 30 you’ll have an AI assistant that knows who you are, remembers you between conversations, answers the way you like, and has already done something genuinely useful for you. Not “one day” useful… today useful.

The half hour, at a glance:

• 5 min - sign up and grab the apps
• 10 min - the settings that make it yours
• 5 min - break the ice
• 5 min - learn how to ask (the actual skill)
• 5 min - give it a real job (three to pick from)
• then - one rule, one habit, done

First, a two-minute pep talk

You cannot break it. There’s no wrong button, no settings maze, no “oops I deleted the internet”. It’s a chat window. You type, it answers, and if you don’t like the answer you just… say so. Nobody’s watching you be new at this, and the people who seem ahead of you mostly just started typing sooner. That’s the whole gap.

First 5 minutes: Get Claude

There are lots of AI tools. Ignore them all for now - you only need one, and my pick is Claude. It’s the one I use every single day to run my business, it’s the one I teach with, and it’s the best writer of the lot by a mile. Also it’s free.

1. Go to claude.ai and sign up with your email (or Google account).
2. While you’re there, grab the desktop app from claude.ai/download. You don’t need it - the browser works fine - but honestly it’s where I live all day: one click in your dock or taskbar instead of hunting through seventeen tabs, and it makes Claude feel like part of your computer rather than another website.
3. The phone app’s worth having too (App Store or Google Play - and it’s where the lovely talk-instead-of-type voice mode lives).

That was the whole “technical setup”. Told you.

Next 10 minutes: Set it up properly (the bit everyone skips)

This is what separates “I tried AI once, it was a bit meh” from “how does it just get me?”. Ten minutes here pays you back forever. Open Settings (click your initials or the menu icon) and do these four things:

1. Introduce yourself. In the Profile section, give Claude your name and what you do, then find the box for describing how you’d like it to respond. Fill in the blanks and paste this in:

Nick this one, fill in the blanks:

I'm [name], and I [what you do - job, business, home life, whatever fills your days]. I'm fairly new to AI, so: explain things in plain language, no jargon - if a technical term is unavoidable, define it in brackets - and keep answers short unless I ask for more detail. Use British English. Talk to me like a helpful friend, not a manual. If I ask for something and you need more information to do it well, ask me questions instead of guessing.

That last sentence is quietly the most powerful thing in this whole guide. It turns Claude from a machine that guesses into an assistant that checks.

2. Turn on memory. In Settings → Capabilities, switch on memory and “search and reference chats” if they aren’t already. This is the difference between an assistant with amnesia and one that remembers your business, your projects and your preferences between conversations. It’s free for everyone now, and it’s the single most “how does it know that?!” feature Claude has.

3. Decide your privacy setting. In Settings → Privacy (sometimes called Data Controls), you’ll find a toggle about helping improve Claude - it controls whether your chats can be used to train future models. There’s no wrong answer, but it should be your answer, not a default you never saw. If the idea makes you uneasy, turn it off and move on with your day.

4. If you’ve used ChatGPT before, bring your history with you. Ask ChatGPT:

Nick this one (paste into ChatGPT):

Give me a concise, structured summary of the main things you've learned about me from our chats: my work, my goals, my preferences, and how I like to communicate. Group it by theme.

Then copy its answer, come back to Claude, and say: “Update your memory about me with this” and paste it in. You’ve just skipped the getting-to-know-you phase entirely.

And that’s genuinely all the setup you need today. There’s more under the bonnet - Styles, Projects, voice mode, the model picker - but none of it is day-one stuff, and the defaults are good. (When you want the full guided setup, session by session with a playbook for your exact job, that’s what The Claude Kickstart is for - coming soon.) One tip: menus move around occasionally, so if a toggle isn’t where I said… poke about, it’ll be nearby. Nothing bites.

5 minutes: Break the ice

Now the fun bit. Don’t sit staring at the empty box wondering what clever thing to type - make Claude do the work. Nick the prompt below, pop it in, press enter:

Nick this one:

I'm brand new to AI and you're the first one I've properly tried. Ask me five quick questions about my week - work, home, whatever - and then suggest three small, specific things you could take off my plate. Don't do them yet, just show me the menu.

Notice what just happened: it asked about your life and offered to fix your problems. That’s the moment most people get hooked… it’s not a search engine, it’s help.

5 minutes: How to actually talk to it

Here’s the skill under everything else, and it’s not “prompt engineering”… it’s just being a decent boss to a very keen new assistant. A good ask has three parts:

• The situation - what’s going on and who’s involved. “I run a small cleaning business and a customer has cancelled last minute for the third time.”
• The job - what you actually want, with a proper verb. “Draft a message that’s firm about my cancellation policy but keeps things friendly.”
• The shape - how you want it served. “Keep it under 100 words, no corporate speak.”

Compare that to typing “help me with a difficult customer” and you can see why one gets gold and the other gets waffle.

And the real secret, the one nobody puts on a poster: the first answer is a first draft, not a verdict. Haggle with it. “Warmer.” “Shorter.” “You’ve made me sound like a solicitor, try again.” Each nudge takes five seconds and the third version is nearly always the keeper. People who are “good at AI” aren’t writing magic prompts… they’re just willing to say “not quite, again” a couple of times.

Last 5 minutes: Give it a real job

Three to choose from - pick whichever matches your actual day. One is plenty for five minutes; the other two will keep for tomorrow. The point isn’t the demo, it’s watching it do your thing.

The email untangler:

I'm going to paste in an email. Tell me in two sentences what they actually want, list anything they're asking me to do (with deadlines if mentioned), then draft a reply that's friendly but brief. Here's the email: [paste it]

The dinner sorter:

Here's what's in my fridge and cupboard: [list roughly what you've got]. Give me three dinner ideas I can make with this, each under 30 minutes, and a mini shopping list for anything small I'm missing.

The confusing-letter translator:

I'm pasting in a letter I find confusing. Give me: 1) What it's actually saying, in plain language, three sentences max. 2) Whether it needs anything from me - if so, exactly what and by when. 3) What happens if I ignore it. 4) Anything worth double-checking with the sender directly rather than taking your reading of it. Don't guess: if something in the letter is ambiguous, say so. Here it is: [paste it]

Whichever one you pick, odds are it’ll genuinely surprise you - the moment it does something in seconds that usually eats twenty minutes. That moment is the point of the exercise.

Before you go: One rule and one habit

The rule: never paste in passwords, bank details, or other people’s private information. Treat it like a very capable stranger - brilliant with your to-do list, not your PIN. (There’s a whole guide on this - The 5 Things You Should Never Paste Into AI - because it deserves one.)

The habit: tomorrow, before you Google something, write something from scratch, or sit staring at a task you’re dreading… ask Claude first. That’s it. That one reflex, repeated, is how people get “good at AI”. There’s no other secret, I promise.

The jargon, translated

Eight words you’ll bump into, in plain English, so nobody can make you feel daft:

• Prompt - the thing you type. That’s it. Every message you send is “a prompt”.
• Model - the brain behind the chat. Claude, ChatGPT and Gemini are different models, like different makes of car - and Claude itself comes in a few versions, which is what that picker at the bottom is about.
• Hallucination - when AI confidently makes something up. It happens… which is why you double-check the important stuff.
• Context - everything the AI can “see” in your current conversation. Long chat = lots of context. New chat = clean slate (unless memory’s on… which is why you turned it on).
• Custom instructions - the tell-it-about-yourself-once settings you filled in earlier. Congratulations, you’ve already done the thing with the fancy name.
• Memory - Claude’s notes about you that carry between conversations. You can ask it what it remembers, and tell it to forget things, any time.
• Projects - folders for your chats, each with its own instructions and files. Lovely once one area of your life (the business, the house move) deserves its own space.
• Artifacts - when Claude opens a side panel and builds you an actual thing - a document, a plan, even a little web page - instead of just chatting. Don’t be alarmed, it’s showing off.

If something felt off

• It got a fact wrong. That’s the hallucination thing - like a keen new assistant on their first week. Double-check anything important. Doesn’t make it useless, makes it human-adjacent.
• It sounded robotic. Tell it! “Again, but how a friend would say it” works wonders. It takes feedback better than most people.
• It stopped and told you to come back later. You’ve bumped into the free plan’s usage limit - it resets after a few hours, nothing’s broken, and you haven’t been banned. It’s actually a decent sign: it means you were properly using it. If you hit it often, that’s when the paid plan starts earning its keep.
• You froze at the empty box. Completely normal. Steal prompts instead of writing them - that’s exactly what The 10 Prompts That Do Your Admin is for.

Right, go on then

That’s your half hour: signed up, properly set up, memory on, introduced, taught how to ask, a real job done, one rule, one habit, and the jargon defused. You’re no longer behind… you’re someone who uses AI, with the settings to prove it. Go you. ✨

Next up when you’re ready: The 10 Prompts That Do Your Admin - ten copy-paste jobs for the boring stuff.