⚙️ The 5 Settings That Make AI Actually Yours

Two people can type the exact same question into Claude and get completely different answers. Not because one of them is cleverer, or wrote some magic prompt. Because one of them spent ten minutes in the settings, and the other never opened them.

I think of the default experience as the showroom problem. A furniture showroom is arranged for nobody in particular - beige sofa, neutral cushions, inoffensive lamp. It’s fine. Nobody’s ever loved it. Out of the box, AI is the same: tuned for the average of everyone, and the average of everyone is no one. Certainly not you, with your specific job, your specific week, and your specific allergy to corporate waffle.

So this guide does one narrow thing deeply: the five settings that turn the showroom into your living room, and why each one earns its place. I call them the Five Dials. For each dial you get two lines - where it lives, and the one move that matters - then the reason it changes everything.

Dial 1: The introduction

Find it: Settings → Profile (click your initials or the menu icon). Why bother: tell Claude who you are and how to talk to you, once, in writing.

Here’s why this dial comes first. Every time you ask an untuned AI a question, it’s answering a stranger. It has to guess whether you want three lines or three pages, whether “budget” means a household or a business, whether you’d rather be reassured or given it straight. Guessing produces beige.

The Profile section fixes that permanently. There’s a spot for your name, what you do, and how you’d like Claude to respond. Fill in the blanks below and paste it into that last box:

Nick this one, fill in the blanks:

About me: I'm [NAME]. My days are mostly [ONE LINE ON YOUR WORK OR LIFE - "running a small campsite in Wales", "two kids and a part-time admin job", "job hunting after redundancy"].

How to respond to me: use British English. Give me the short answer first and offer more detail after, rather than the essay up front. No jargon unless you explain it in the same sentence. If my request is vague, ask me one or two questions before you start instead of producing a polite guess. And if something I've asked for is a bad idea, say so and tell me why - I'd rather hear it from you.

Say you’re a mobile hairdresser. Yours might read: “I’m Dee, I do mobile hairdressing around Stockport, mostly older clients, and I do my own bookings and accounts.” Now ask “help me with my prices” and instead of a lecture on pricing strategy, Claude asks about your travel time, your regulars, and whether you’ve raised prices since your products went up. Same question, different assistant.

The bit everyone misses: those last two sentences in the paste-in. “Ask before guessing” and “tell me when it’s a bad idea” are the antidote to AI’s most annoying habit, which is enthusiastically agreeing with whatever you say. You want an assistant, not a mirror.

Dial 2: The memory switch

Find it: Settings → Capabilities. Why bother: turn on memory, and turn on “search and reference chats” while you’re in there.

Without memory, every conversation starts from zero. You explain your business on Monday, and on Wednesday you’re explaining it again, like a very polite Groundhog Day. With memory on, conversations compound. Mention in week one that you’re renovating the spare room on a tight budget, and in week four, when you ask about flooring, it already knows the room, the budget, and that you hate laminate.

That compounding is the entire reason regulars get such good answers. Their tenth conversation isn’t starting where your first one is. And it’s free for everyone, which still surprises people.

The non-obvious bit: memory only contains things you’ve actually said to Claude. It isn’t reading your email or watching your screen. Which also means it can be wrong in mundane ways - if you change jobs and never mention it, it’ll cheerfully keep assuming the old one. Wrong memories can be corrected by just telling it: “forget that, here’s the update.”

Dial 3: The privacy decision

Find it: Settings → Privacy (sometimes labelled Data Controls). Why bother: find the toggle about using your chats to improve the models, and set it on purpose.

This dial doesn’t change Claude’s answers at all. It changes something quieter: how much you trust the thing you’re typing into.

A default you never chose has a way of nagging at you. People who feel vaguely uneasy about “where their stuff goes” hold back - they paste in pretend problems instead of real ones, half-use the tool, and then conclude AI is a bit useless. Of course it is, on a diet of pretend problems.

There’s no wrong position for this toggle. Leave it on, switch it off - both are respectable. The only wrong state is “never looked”. Thirty seconds, decide, done, and now you can use the thing wholeheartedly. That’s the real feature you just switched on.

Dial 4: The front door

Find it: claude.ai/download - the desktop app. Why bother: put Claude one click away instead of one browser-tab-safari away.

This one looks like it’s not a setting at all. It’s the setting that decides whether the other four ever get used.

New habits lose to old habits unless they’re physically closer. If asking Claude means finding the browser, finding the tab, getting distracted by the other forty tabs… you’ll just Google it like it’s 2019, or worse, do the task by hand. The desktop app sits in your dock or taskbar like a colleague at the next desk, and that tiny drop in friction is the difference between “I keep meaning to use AI” and actually using it. It’s where I live all day - my whole business runs through that one window.

While you’re at it, the phone app is worth the download too, mostly for voice mode: talking through a problem on the school run beats typing it at midnight.

Dial 5: The voice

Find it: the Styles menu, in the chat box itself. Why bother: if you keep correcting the same thing, stop correcting and make it a style.

Here’s the test. If you’ve typed “shorter” or “less formal” or “stop being so enthusiastic” more than three times this week, you’re not prompting anymore - you’re manually applying a setting, over and over, by hand.

Styles let you set the tone once. Pick one of the built-in ones or create your own (“brief and direct, no pep, no bullet points unless I ask”) and Claude writes that way in every chat where it’s selected, without being reminded.

The mistake to avoid: don’t build five elaborate styles on day one. You’ll never use them, like the fancy stitches on a sewing machine. One style that matches how you actually like to be spoken to covers 90% of life. And note the division of labour - your Profile tells Claude who it’s talking to, a Style tells it how to sound. They stack nicely.

Where this stops working

Being honest with you:

• Settings sharpen answers - they don’t come up with questions. A perfectly tuned Claude still sits there silently until you ask it something. Knowing what to ask is a separate skill, and it comes from use, not toggles.
• Nothing here makes it factually infallible. A personalised wrong answer is still wrong. Keep double-checking anything that matters.
• This is deliberately not the whole tour. Projects, the model picker, Artifacts and friends are all real, and all fine on their defaults for now. Five dials is the point, not the ceiling.

The one move

Dial 1. Two minutes in the Profile settings improves every single conversation you’ll ever have, forever, with no further effort. It is the best-value two minutes in all of AI.

Once memory’s been on for a few weeks, it’s worth knowing exactly what’s built up in there - that’s what What Claude Remembers About You (And How To Edit It) is for. And when you want the full guided setup with a playbook for your exact job, that’s the job of The Claude Kickstart (coming soon). ✨