Divyansh Sharma

How to Build an AI-Ready Second Brain, here is how I have done.

3 min read

A system that remembers when you don’t

A while ago reading building a second brain by Tiago Forte.

I was like yes a place where you keep everything stored at one place.

So convenient

  • someone needs a sign, here you go

  • someone needs your portfolio, here you go

  • you need a document you last used in December 23 BC

it is there too

Why it matters more now, with AI

in the age of AI, agentic AI, we now are at a place where it is extremely important, you need a skill.md file to feed into AI, or you need it to check your GitHub projects, or a folder which contains details about everything you are working on.

If you have this set properly, it can be sorted out really good, to a point where, you will be like can it do this? and you can create a custom solution for it.

You want to remember everything claude did on 13th July 2026, you can ask it to write a script, that it updates it day wise.

The point is you can be disgustingly well organized, and that gives you an unfair advantage.

Imagine you live in a house, it has everything in its place, and suddenly the clock is out of battery.

you know exactly where to look.

Now imagine taking that power and giving it to an AI, it has the exact project at its disposal, it has the exact context, it has everything it needs to build what you need.

Leaving the AI agent to the side, you yourself can use it hyper well,

The structure of my second brain

This is the structure of my second brain

  1. Projects (everything I am working on currently)

  2. Areas (everything I am working on for my entire life)

  3. Resources (everything I can need)

  4. Archives (Past projects and documents)

The book Building a second brain dives deep in this. Worth checking out.

Two folders I added: Ideas and Heart

I added Ideas and Heart to this

Ideas are the ideas those I would/or would not execute

Heart contains private things, things you would not want anywhere

-API keys and stuff.

Keep it separate and in a different place

Give your AI an index

How can you take it to some other level?

You can create a separate index, which basically tells your LLM where it can find what, so in a way it never loses context and you do not have to manually copy paste, this system works wonders if you are building a big project and keep running out of context, you can also trigger a script, that logs when it is doing anything so it knows what was edited and when and like me if you are using more than 1 simultaneously it knows if some other agent is working on anything so it does not touch what is running.

And the thing is you can get super creative with it.

You can organise it how you use it, and infact that is how you should from where you use it, in the book the same concept is explained it depth, keeping it project wise, where it is used, unlike a library where you are organising domain wise, it should be more like

there might be a project where you are using science, code, some research you did on sleep, instead of categorising it on the basis of subject, but rather the project, so it all stays in one place and if you are using one thing at 2 places you can link it there, so you just have to look at one place and be done with it instead of you having to search every time.

I am still learning these super curious on how you use yours, would love to hear you and learn how I can make it better.

Thank you for reading