a free 7-page guide from soft tech

First 5 Steps to
Setting Up Claude

Set it up once. Stop getting generic answers forever.

"It was never broken. It just didn't know you yet."

Start here
You're not bad at this. Your Claude is just a stranger.

Here's what happens to almost everyone. You hear Claude is the tool. You open it. You ask it something real. And it hands back a paragraph so flat you could have written it half-asleep. So you decide the problem is you. You're "not a tech person." You close the tab.

Let me stop you there. The problem was never you, and it was never the tool. The problem is that you never set it up. Out of the box, Claude is a stranger. It doesn't know your work, your voice, or how you want things done. A stranger gives generic answers. That's all a stranger can do.

Setting it up is how you fix that. Not with code. Not with settings buried six menus deep. You put five things in place, once, and Claude works better every single time after that.

This is the difference between people who say Claude is overhyped and people who say they can't work without it. The second group set it up. That's the only difference.

Do these five in order. Each one takes a few minutes and you only do it once.

Step one
Give it a home that knows you
Sets up: context, so it stops starting from zero.

Every time you open a fresh chat, you're meeting a stranger again. It forgets you the second you close it. So you re-explain who you are, over and over, and you still get bland results.

Give it a home instead. A project, with one short file about you. You build this once and every chat inside that project starts already knowing the basics.

Tell it what you do. Who you serve. What you're working on right now. What matters to you. Save it. Now it has context before you type a single word.

"I want you to know me before we work together. Ask me 8 questions about my work, my goals, and how I like to communicate. Then write a short 'about me' summary I can save to this project so you remember it every time."

That's the foundation. Everything else sits on top of it.

Step two
Hand it your voice
Sets up: writing that sounds like you, automatically.

You know the sound. "I am thrilled to announce." "In today's fast-paced world." Stiff, formal, nothing like how you actually talk. Claude writes that way by default because it's guessing at a voice. Set yours up once and it stops guessing.

Give it a sample of your real writing and tell it what to learn. Save what it gives you back into your project, right next to your "about me" file.

"Here is something I wrote. [paste your writing.] Study how I write. My sentence length, my tone, the words I use and the ones I'd never use. Write me a short style guide based on this, and save it so you write in my voice every time."

From now on, every draft starts in your voice. Not a press release. You.

Step three
Set its ground rules
Sets up: how it behaves, so you stop correcting the same things.

Notice the things you fix over and over. It writes too long. It uses jargon you hate. It dives in before it understands the question. You can stop fixing those one at a time and set them as standing rules instead.

These are your default instructions. Claude follows them in every chat without being reminded. You set them once.

Think about your three biggest "ugh, not like that" moments and turn each into a rule.

Add your own. Now it shows up the way you want from the first message, every time.

Step four
Connect it to your real stuff
Sets up: access to your actual work, so you stop copying and pasting.

Most people describe their work to Claude. They summarize the long email. They explain what the spreadsheet "kind of shows." That middle step is slow, and you lose detail every time you translate.

Connect Claude to where your real work already lives instead. Your email. Your calendar. Your documents. You connect it once, and after that it can work from the real thing instead of your description of it.

Setup is a few clicks, not code. You approve which tools it can see. You stay in control of what's connected.

  • "Read the email from this client and draft three replies. One warm, one short, one that buys me time."
  • "Look at my calendar this week and tell me where I have two free hours."

No copying. No pasting. It already has what it needs.

Step five
Save what works as a Skill
Sets up: your repeat tasks, so you never rebuild them.

The first time you get Claude to do something well, that whole effort, the prompt, the format, the rules, is worth keeping. Most people throw it away and rebuild it from scratch next week. You don't have to.

A Skill is a saved instruction set you trigger with a slash command. You teach Claude how you want a task done once. The brief, the format, the rules, the examples. You save it as a Skill. After that it's one command, and every instruction is already baked in.

You build it once. You run it for the next year.

"Turn what we just did into a Skill I can run with a command. Include the format, the rules, and a blank for the parts that change each time."

Your weekly captions. Your client follow-ups. Your meeting notes. Each one becomes one command instead of one more hour.

That's the whole setup
Five things. One stranger, turned into a tool that works for you.

Look at what you just built.

A home that knows you. Your voice, baked in. Ground rules it actually follows. A connection to your real work. And your repeat tasks saved as Skills you run with one command.

That's the difference between "Claude is overhyped" and "I don't know how I worked without it." None of it was technical. All of it was setup you only do once.

Do these five this week. One a day if that's easier. By Friday you'll have a Claude that actually feels like yours.

When you're ready to go deeper, Getting Started with Claude walks you through all of it with me, step by step. (Coming soon.)