There’s a version of knowledge management that looks great in productivity blogs and falls apart by week three. You’ve probably tried it. The Second Brain concept, capturing everything, linking ideas, keeping contact records fresh, is compelling in theory. In practice, it costs roughly an hour a day to maintain properly. And most people quietly abandon it.
Our CEO Marek Elznic was one of those people. Except instead of giving up on the idea, he redesigned the problem.
The talk that started the conversation
Last month, Marek was invited to speak at Orbitant’s Knowledge Sharing session. Orbitant is a senior engineering consultancy with a genuine learning culture built in: weekly knowledge sessions, documented technical decisions, teams that actively challenge each other’s thinking. The kind of environment where ideas get pressure-tested, not just applauded.
His topic: personal knowledge management in the age of AI. Specifically, how he stopped waiting to become the kind of person who would spend an hour a day on note-taking, and built a system that doesn’t need him to.
The real problem with Second Brains
The Second Brain methodology is built on a simple premise: externalize your thinking so your mind is free to connect ideas rather than store them. It works. The problem isn’t the concept, it’s the maintenance tax.
Writing structured notes takes time. Linking ideas across files takes time. Keeping contact records and project logs current takes time. Multiply that across weeks and months, and you’re looking at a part-time job for a system that was supposed to make your work easier.
Marek’s honest admission at the Orbitant session: he wanted that compounding knowledge system. He just accepted he was never going to put in that time.
Claude as the maintenance layer
So he built something different. The vault still lives in Obsidian, a local, markdown-based knowledge tool that’s become popular among developers and knowledge workers for its flexibility and linking system. But the upkeep is handled by Claude.
Marek’s setup runs a nightly routine: Claude writes the notes, links the files, updates the records, and keeps everything current, automatically, without him touching it. The knowledge compounds. The effort doesn’t.
It’s a small but important shift in how to think about AI in your workflow. Not as a search tool you query when you need something, but as an active layer running in the background, doing the unglamorous work that most systems require a human to do.
Why this matters beyond productivity
The broader point Marek made, and the one that stuck, is about compounding. A knowledge system only gets valuable through the connections that accumulate over time. Most people never reach that stage because the maintenance collapses before the compounding kicks in.
Delegating that layer to AI isn’t cutting a corner. It’s what finally makes the long-term payoff reachable.