Let’s be real, it was months of pull requests, incidents and decisions that all felt significant at the time, but not everything got written down or it’s not that easy to find.

Have you ever heard of an AI second brain? The idea is that you keep a record outside your head so you stop relying on remembering. This is not a new concept and it’s been around for some time now. However, AI came to grow this concept a lot more.

What AI adds is that the record can answer questions, so you no longer have to go hunting through it, which was the hardest part of every system people tried before.

What exactly is an AI second brain?

The term AI second brain covers three different things:

• The original version is a notes archive you organise yourself, popularised by Tiago Forte’s 2022 book Building a Second Brain, which set out a four-step method: capture, organise, distil, express. It works but it’s a lot of manual labour.
• The second is when your notes and documents sit in a tool like Notion or Obsidian, and an assistant answers questions from them on request.
• The third is the most common nowadays: a memory an AI assistant reads from and writes to while you work. You talk/write, it stores, and next time you ask something it already has your context.

Capturing information was not the real problem in the last few years, but the retrieval of that information was what was harder to do and AI gave us a faster way to solve that problem.

How to build an AI second brain

Finding a system that works for you is the easiest part. The habit you need to keep it updated is the challenge. But first things first:

• Pick one place to put your system and only one. It can be the memory inside the assistant you already use, a plain Markdown folder an assistant can read, or a notes app with an assistant attached.
• Keep each note small enough that you will still write it on a bad day. The weeks most worth recording tend to be the busiest ones, so anything that feels like a task gets skipped. Two lines at the end of something is plenty, and dictating them into your phone can be a solution instead of typing.
• Write entries that answer a future question. “Fixed the queue thing” hardly helps you. “Queue backlog was hitting 40k at peak, moved to batched consumers, down to under 500” however, is the same note but you can use this information easily in the future.
• Go back to it once a month. Ask it what you did and what it thinks you learned. This takes about ten minutes and leaves you with something you can use at review time. It is also how you spot a month where nothing got written down, while you can still remember it.

What to be careful with

Handing cognitive work to AI can have an impact on your brain and you should be careful with it.

Michael Gerlich’s 2025 study of 666 UK adults, published in Societies, found heavier AI use associated with weaker critical thinking. The MIT Media Lab ran an EEG study on 54 people writing essays, and the group using an assistant showed the weakest brain connectivity of the three, with most unable to quote a sentence from work they had finished minutes earlier. 

The International AI Safety Report 2026 describes this field as nascent while citing a study where clinicians’ ability to spot tumours without AI dropped by 6% three months after the AI arrived. 

Education researchers, Lodge and Loble argue that offloading is neither good nor bad in itself, and that what decides it is what happens to the capacity you free up. If you question the results you are given and if you work on top of AI, your brain should be fine.

A second brain is still your brain. It holds what you did, and the thinking and processes are still yours.  

You have done more than you can remember, and that is the whole problem. Start with one note after your next deploy, and in a year you will have a whole system.