Hi there,

I want to give you something very important this week — the full Claude setup that now runs basically every piece of writing and planning in my work.

This is the most useful thing I've shared all year, and I'm not exaggerating. If you've been following along on the podcast or in past issues and wondering what the AI workflow I keep referencing actually looks like under the hood — this is it.

Most people use Claude like a search engine. They type a question, get an answer, close the tab.

In the beginning I did this too.

Then I spent one afternoon setting it up properly. I uploaded a few files about how I think and how I work. I told it what I'm building, what I sound like, what I'm bad at, and what I want it to never do.

The next morning, I wrote a LinkedIn post in 6 minutes that used to take me 40.

Not because Claude got smarter. Because Claude finally knew me.

It's not a special model and it's not some magic prompt. It's a setup that takes about 2 hours, costs $20/month, and means you stop re-explaining yourself every single
time you open a new chat.

Below is the full thing — every file, every prompt, every shortcut.
This is the longest email I've ever sent, so save it. You'll come back
to it.

A folder of files about you, inside Claude, that the model reads before answering anything.

The files cover:

That's really the whole secret.

With these files in place, your prompts shrink from 800 words to 12. You can say "write a LinkedIn post about yesterday's podcast" and Claude already knows your voice, your audience, your format, your banned words, your call-to-action style. The first draft comes back usable.

Without these files, every chat starts at zero. You explain who you are. You explain the project. You correct the tone four times. You give up and rewrite it yourself.

The 2 hours you're about to spend is the difference between those two worlds. And honestly, this is the highest-leverage time investment I've made in my work this year.

I use all of those tools too. They're great at different things — I still grab ChatGPT for image generation and Perplexity for real-time research.

But for this specific setup, Claude wins on three things that matter a lot:

If you don't have Claude Pro yet, it's $20/month. You do need the paid version for this — Projects don't exist on the free plan.

This is something I'd do on a Saturday morning with coffee and a clear head. It's not hard, but it's worth giving it your full attention because you'll only do this once.

Head to claude.ai, then go to Sidebar → Projects → Create project.

Name your Project after one thing you do a lot — "LinkedIn," "Client emails," "Investor updates." I'd start with the task you do every single week and want to be 5x faster at. (You can always add more Projects later.)

Here's the part I love most: you don't write this yourself. You have Claude interview you.

Open a new chat in your Project and paste this prompt:

Answer everything honestly. Don't write what you wish were true — write what's actually true, including the messy stuff.

When Claude finishes, copy the document it generates and upload it to your Project as about-me.md. This is the most important file in the whole setup, so spend the 30 minutes. Future you will thank present you.

Same approach for this one — don't describe your voice in the abstract. Let Claude figure it out from the work.

Open another new chat in the same Project and paste this:

Paste in 3 things you've written that sounded most like you. A LinkedIn post that hit. An email to a friend. A caption that got real engagement.

Answer the follow-up questions Claude asks — they're usually really good. Save the document, upload to your Project as my-voice.md.

This is the file that stops Claude from making your writing sound like Claude.

I built mine after watching my own newsletter drafts come back full of em dashes, "not just X but Y," and the word "navigate." I made a list of every AI tic I could spot and told Claude: never do these.

It works really well. Mine is public — drop it into your Project as a starting point, then add your own banned words and phrases.

→ Open the AI Second Brain folder — grab anti-ai-writing.md from the "Files to upload to Claude" subfolder (use md.file, it consumes less tokens, but pdf. file is also available so you can read it yourself).

Inside your Claude Project there's a field called "Project instructions." This is your system prompt — it runs before every single chat in this Project.

Now pull project-instructions-template.md from the same folder. Fill in the [BRACKETS] with your details, paste it into your Project Instructions field, and save.

Take the 3 things you write most often.

For me right now it's newsletter intros, LinkedIn posts, and emails to my podcast guests. For you it might be investor updates, sales emails, or deck outlines. Pick what's true for your week.

For each one, save your best example — the version that actually worked in real life — as its own file:

This is the step most people skip, and I really don't want you to skip it.

Drop in everything that shows Claude where this project currently stands. Real files, numbers. The messy reality.

If your Project is LinkedIn → upload your last analytics screenshot, upload a CSV of your recent posts and their performance, paste in 5 of your best-performing posts as a doc.

If your Project is Sales → upload your pipeline, your win/loss data, a few of your closed deals.

And then write one line at the top of your Project Instructions: your big goal for this year. Something like:

That one line changes everything. Now when you ask Claude "should I post this?" — it's not guessing what good means. It's checking your post against your real goal and your real numbers.

The more context you give, the less guessing Claude does. Stats, screenshots, CSVs, past work — pile it on.

Everything I just walked you through — files, voice, project context — gets you to roughly the same place as a smart assistant who knows you. That's huge, and most people would be thrilled to stop there.

But the actual unlock is the next step, and it's the thing I'm most excited about right now: building a closed feedback loop with your AI.

What I mean is this. Every time you give feedback to your team, every time you edit something Claude made, every time you say "no, change this" — that information has to find its way back into Claude. Otherwise Claude keeps making the same kinds of mistakes, and you keep correcting them, and nothing compounds.

I'm still figuring out the cleanest way to do this. But here's what I'm experimenting with right now, in case it sparks something for you:

This is the most important shift, and the one I think will define the next year of working with AI. The speed of your decision-making goes up dramatically the second Claude starts predicting your judgment instead of asking for it every time.

I'll write more about this as I work it out. For now: start recording, start sharing your edits with Claude, start moving feedback into AI-readable channels. Even messy first attempts beat nothing.

I put every template I mentioned into one Google Drive folder.

Don't try to read everything at once. Open START_HERE.pdf first — it's a one-page intro that tells you exactly which file to use when. Takes 2 minutes.

Everything else lives in the same folder, and you'll know what to do with each one after you read START_HERE.

→ Open the AI Second Brain starter pack →

That's it. About 100 minutes if you stay focused.

You now have a Claude Project that knows who you are, how you write, what you're working toward, what your real numbers look like, and what good output looks like for you.

From this point forward, your prompts can be 12 words long and they'll come back usable. It's such a satisfying feeling, honestly.

Have a great Sunday!!

Marina 💜