The problem the second brain promises to solve
McKinsey calculated that 20% of the working week is spent searching for information. Not producing it — searching for it. Re-reading emails, reconstructing context, finding that document you were sure you had saved somewhere.
The problem is not laziness. It is that the human mind is optimised for thinking, not for remembering. You can have years of experience in a field, have read hundreds of books, attended thousands of meetings — and most of that wealth remains trapped in your head. Or worse, it gets lost.
The second brain idea starts here: build an external system that remembers on your behalf. Not a passive archive where you dump things at random, but a structure that reflects the way you think and work. Augmented memory that gives you back what you know at the moment you need it.
It sounds ambitious. And for years it remained, at least in part, an unfulfilled promise.
What a digital second brain is
A digital second brain is an external system that stores, organises and connects what you learn over time. It is not a list of links, not a folder of downloaded files. It is a network of connected notes that mirrors your mental structure.
The concept was systematised by Tiago Forte with the CODE method: Capture (capture relevant information), Organise (make it accessible), Distil (extract the essence), Express (use what you know to create something new). A continuous cycle, not a one-off procedure.
The preferred tool among second brain practitioners is Obsidian. The reason is simple: it works locally, uses markdown files, and does not depend on any vendor. Your data stays yours — not on an external server, not tied to a subscription. And the markdown format is readable by any system, including AI.
The difference from a simple collection of notes lies in the connections. In Obsidian every note can link to other notes. Over time a map of your knowledge emerges — ideas connect, patterns become visible, relationships between different concepts surface on their own.
Why the pre-AI version did not hold up over time
Anyone who tried to build a second brain before AI entered the workflow knows the problem well. You start with enthusiasm, create the structure, populate the first notes. Then, after a few months, the vault has become a chaotic repository you use less and less.
The paradox of the classic second brain: the more it grows, the less usable it becomes. The more notes you add, the harder it is to find what you are looking for. The categories you created are no longer enough. Notes age without updates. Keeping everything in order takes time — and that time is often not available.
The real problem was the relationship between maintenance cost and value produced. You spent more time organising than using. The vault became a project to keep going, not a tool in service of the work.
Many people gave up after a few attempts. Not because the idea was wrong, but because maintenance required almost unreasonable discipline. Without someone — or something — to help keep everything updated and connected, the system collapsed under its own weight.
Want to bring the second brain concept to your company?
30 minutes to discuss your specific case.
What changes with Claude
The conceptual shift is this: with Claude, the vault is no longer a passive archive. It becomes an active cognitive system.
The typical setup: Claude Code opened inside the Obsidian vault folder. Claude has direct access to all the notes in markdown format and can read, write, connect and synthesise content on demand. You can ask it "what do I know about this client?" or "what decisions did we make about this technology in the last six months?" — and it answers by drawing directly from your notes.
The key is the CLAUDE.md file. It is a text file that lives in the root of the vault and contains the context you want Claude to always have in mind: who you are, what you work on, how you prefer to receive information, what conventions you use. Ten lines that change everything — Claude "knows" you from the very first message.
Andrej Karpathy called this approach an "LLM Wiki": a markdown archive structured so that a language model can reason over it effectively. You do not need a vector database, you do not need complex infrastructure. A well-organised markdown file structure is all it takes.
With scheduled agents you can go further: scripts that update the vault overnight, integrate new sources, flag outdated information. The system starts to run itself.
Who uses it and what for
The second brain with Claude is not a tool for automation nerds. It serves people who manage a lot of information and want to do so without losing their minds.
Managers use it to track decisions, meetings and projects. No more "where did I write that?" — instead a natural language query that returns the full context.
Consultants and professionals use it for research, client management and methodologies. Every client has their own note. Every methodology is documented. Over time a knowledge asset accumulates that does not depend on personal memory.
Entrepreneurs use it for strategy, ideas and pipeline management. The second brain becomes the place where ideas mature before becoming decisions.
Technical teams use it for documentation, architectures and project decisions. Fewer "why did we do it this way?" questions in code reviews, more context available for anyone who joins the project.
In all these cases the common thread is the same: turning individual knowledge into something accessible, up-to-date and useful — without maintenance becoming a second job.
How to get started: the minimum viable setup
You do not need to be technical. You do not need to know how to code. You just need the willingness to build a habit.
The starting point: install Obsidian (free), create a vault, open Claude Code in the same folder. In under thirty minutes you have a working environment.
The most important step is creating the CLAUDE.md file in the root of the vault. Just a text document with your context: who you are, what you do, what you are working on, how you prefer to receive answers. Ten lines, no more. That becomes the starting point of every conversation with Claude.
It is not a project to finish. It is a system that grows with you. The first week you add the most important notes. The first month you build the connections. After three months you have something that genuinely delivers value.
If you are thinking of scaling it to the company level — a shared knowledge base, agents that automatically update documentation, integrations with existing systems — that is where Maverick AI comes in. We design and implement knowledge management systems with Claude, from personal vaults to organisation-wide knowledge bases.