PROMPT CHAMPS
Video guides / Claude

Claude / Prompt Champs

Claude Projects: Build Your Second Brain

05:47 runtimePublished June 26, 202610 key ideas

You already wrote down the answer. You just can't find it, and it's quietly costing you hours. Three months ago you made a smart call on pricing. You wrote it somewhere: the notes app, a doc, a voice memo, an email…

Watch on YouTube

The complete breakdown

You already wrote down the answer. You just can't find it, and it's quietly costing you hours. Three months ago you made a smart call on pricing. You wrote it somewhere: the notes app, a doc, a voice memo, an email to yourself. Now you need it, you can't find it, so you give up and guess. That's the dirty secret of the second brain everyone tells you to build. A pile of notes you can't search isn't a brain, it's a junk drawer. And here's what makes it worse. When you finally turn to your AI for help, it doesn't know any of it either. So every morning you start over: I run a small bakery, two people, keep it warm and casual, no jargon. You brief a total stranger from scratch, because it forgot you the second you closed the tab. You're losing on both ends. You can't find what you know, and your AI doesn't remember who you are. Most people just accept this. But what if one place held both, your knowledge and your context, and every chat opened already knowing it? You ask, what did I decide about pricing in March, and get it back in seconds, in your own words. That's a Claude Project. Let me show you the ones worth building.

Use this page with the video

Watch the episode above for the visual explanation, then use the notes below to revisit each idea, example, and practical move.

01

A Workspace That Remembers

You open Claude and it's a blank slate that forgot you the moment you last closed the tab. A Project ends that. Click New Project, name it, and give it two things once: custom instructions for who you are and how you want answers, plus a few files it should always know. Every chat inside it then starts already up to speed - no briefing, no re-pasting, no starting from zero.

Set up once, never brief again
02

Your Notes, Searchable

Somewhere in four years of notes is the decision you need right now - you just can't find it. Drop the whole pile into a Notes project: meeting notes, voice memos, book highlights, all of it. Now you don't scroll, you ask. 'What did I decide about pricing in March?' or 'Pull every idea I saved on onboarding.' It answers from your own words in seconds. Your junk drawer is finally a brain.

Ask your own memory anything
03

Make It Write Like You

Ask plain AI to write for you and it sounds like a press release - polished, generic, not you. The fix isn't a better adjective in your prompt, it's examples. Add three of your strongest past pieces to a writing Project and tell it: match the voice in these - short sentences, dry humor, no hype. Now it writes from how you actually sound. Showing beats describing, every time.

Samples beat 'be casual'
04

Give It Rules, Not Reminders

You keep repeating the same corrections - 'don't rewrite the whole thing,' 'only use our policy' - and the next chat ignores them anyway. Put them in the Project's instructions as standing rules. For code: 'show a diff, not the whole file, match our patterns.' For support: 'answer only from these docs; if it's not there, say so.' It obeys your rules without the nagging.

Standing rules it won't forget
05

Make It Cite Its Sources

The scary part of asking AI to summarize a stack of documents is you can't tell what's real and what it invented. So make it prove itself. In a research Project, load your sources and set the rule: answer only from these, cite the document and page for every point, and flag where they disagree. Now you get a summary you can check - against the files it's already holding.

A summary you can verify
06

Turn Notes Into A Tutor

You're three weeks into a course and the textbook explains it every way except the one that clicks. Build a Tutor project from your own syllabus, slides, and readings, then set the rule: teach only from these, quiz me, and explain what I get wrong in plain words. It becomes a tutor that knows exactly where your class is - and where you're stuck.

Your course, as a patient tutor
07

One Project Per Hat

Your day job, side hustle, and personal life all blur into one chat - so Claude offers a corporate tone for a friend's birthday note. Give each its own Project: Day Job, Side Project, Personal, each with its own instructions and files. Switching Projects switches brains cleanly - the right context every time, with nothing leaking from one part of your life into another.

A clean brain for each role
08

Know When To Skip It

Not everything deserves a Project. Spin one up for a single quick question and you've added setup to something that needed none. The test: will you come back to this? Recurring work - your writing, your code, a class you're taking - earns a Project. A one-time 'what's a good gift for a coworker' is just a chat. Build them for what you do again and again.

Recurring work, not one-offs
09

Share It With Your Team

You built a Project that nails your brand voice, and the new hire is still pinging you to ask how to phrase a Slack message. On a Team plan, share it - everyone opens the same setup, same instructions, same playbook. One person tunes it; the whole team inherits the expert. New folks ship on-brand work in their first week instead of their second month.

One expert, the whole team
10

Build Your Command Center

Now combine everything. Build one Command Center Project - your goals, voice, core docs, and standing rules - the place you open for almost every task instead of a blank chat. Then point Claude Code at your repo from that same setup, and the gap between an idea and a shipped thing collapses to almost nothing. Not scattered chats - one brain that runs your whole workflow.

One brain runs everything

Video chapters

More practical AI guides, without the hour-long lecture.

Subscribe on YouTube