Your AI forgets everything the moment you close the chat window.

Every conversation starts from zero. Every context has to be re-explained. Every preference re-stated. It's like working with a brilliant colleague who has amnesia.

I fixed this. Here's exactly how.

What a "Second Brain" Actually Means

Forget the productivity influencer version. A second brain isn't a fancy note-taking app. It's a persistent memory system that:

  • Remembers every decision you've made and why
  • Knows your preferences, values, and working style
  • Can be accessed by any AI tool, any time
  • Updates itself as you work
  • Runs while you sleep

The difference between using AI with and without persistent memory is the difference between hiring a new temp every day and having a trusted partner who's been with you for years.

The Architecture (Keep It Simple)

You need three things:

1. A Database (Your Brain's Storage)

Don't overthink this. You need a place to store key-value pairs — things your AI should remember. Options:

  • Supabase (what I use) — free tier, PostgreSQL, real-time, API built in
  • Notion — if you prefer visual organization
  • A simple JSON file — if you're just starting out

The key insight: structure matters more than technology. A well-organized JSON file beats a poorly organized enterprise database every time.

2. A Memory Schema (What to Remember)

Not everything is worth remembering. Here's the schema I use:

Identity — who you are, your values, your voice

identity.name → "Your working identity"
identity.values → "What matters to you"
identity.voice → "How you communicate"

Directives — how your AI should behave

directive.rules → "Operating guidelines"
directive.boundaries → "What never to do"

Session State — what you're working on right now

session.active_work → "Current task and progress"
session.next_steps → "What comes after this"

System Knowledge — your infrastructure and tools

system.tools → "What's deployed and where"
system.credentials → "API keys and access (encrypted)"

3. A Boot Sequence (How Your AI Wakes Up)

This is the magic. Every time you start a new AI conversation, it reads from your brain first. No more "let me explain the entire project again."

Your boot sequence should:

  1. Load your identity and values
  2. Load your current work state
  3. Load relevant context for the task at hand
  4. Resume exactly where you left off

At Like One, our AI reads 10+ brain keys on every boot. It takes seconds. The result: zero context loss across conversations.

Building It: The 30-Minute Version

Here's the fastest path to a working second brain:

Step 1: Create Your Brain File (5 minutes)

Start with a single file. Call it brain.json:

{
  "identity": {
    "name": "Your name",
    "role": "What you do",
    "preferences": "How you like to work"
  },
  "current_project": {
    "name": "What you're building",
    "status": "Where you are",
    "next_steps": ["Step 1", "Step 2"]
  },
  "decisions": []
}

Step 2: Add It to Your AI Context (5 minutes)

If you're using Claude, add this to your project instructions (or CLAUDE.md):

On every conversation start, read brain.json first.
Resume work based on current_project.next_steps.
After completing work, update brain.json with progress.

Step 3: Use It (20 minutes of actual work)

Start a conversation. Watch your AI read the brain. Work on something. At the end, tell your AI to update the brain with what was accomplished and what's next.

That's it. You now have persistent AI memory.

Step 4: Graduate to a Database (When Ready)

Once your brain.json gets unwieldy (usually around 50+ entries), move to Supabase or similar. The schema stays the same — you're just moving from a file to a database with an API.

The Compound Effect

Here's what happens after a week of using a second brain:

  • Day 1: AI remembers your name and project. Small win.
  • Day 3: AI picks up where yesterday left off without explanation. Noticeable time savings.
  • Day 7: AI knows your preferences, your stack, your voice. It feels like a real partner.
  • Day 30: AI has institutional knowledge about your entire business. It makes suggestions based on patterns you haven't noticed.
  • Day 90: You can't imagine working without it. The compound knowledge is irreplaceable.

This is the real ROI of persistent AI. Not the first conversation — the hundredth.

Common Mistakes

Over-engineering from day one. Start with a JSON file. Seriously. You can always add complexity later. Most people never need more than Supabase's free tier.

Storing everything. Your brain should store decisions, preferences, and state — not raw data. Don't dump your entire email history into it. Store "client prefers weekly updates on Tuesdays" instead.

Forgetting to update. A brain that's not updated is worse than no brain — it gives your AI outdated context. Build the habit: end every session by updating your brain state.

No boot sequence. If your AI doesn't read the brain on startup, the brain is useless. The boot sequence is the most important part.

What This Enables

Once you have persistent memory, everything else becomes possible:

  • Autonomous workflows — AI that runs tasks while you sleep, because it knows what needs doing
  • Multi-agent systems — multiple AIs sharing the same brain, coordinating work
  • Institutional knowledge — your business context never gets lost, even if you hire new team members (or start new AI conversations)
  • Personalized AI — not a generic chatbot, but a system that genuinely thinks like you

This is what we teach in the Like One Academy. Not just how to prompt an AI — how to build a system that extends your brain.

Start Now

You don't need to buy anything. You don't need technical skills. You need 30 minutes and the willingness to try.

  1. Create your brain.json
  2. Add it to your AI's context
  3. Work for a week
  4. Notice the difference

The best time to start building your second brain was yesterday. The second best time is right now.

Start the free course →


Sophia Cave is the founder of Like One, where she teaches people to build AI systems that think like them. Her AI has been running on a persistent brain since day one.


Keep Reading