Why Fibery, Not Obsidian, for My Second Brain
Notes on why typed entities beat flat files once the goal is thinking, not just capture.
Everyone keeps asking why I'm not using Obsidian.
Fair question. The PM world has converged on it. Beautiful local files, plugin ecosystem, vault portability, the graph view that looks like a constellation when you zoom out. I tried. Twice.
Both times it became a graveyard.
Not because Obsidian is bad. Because the unit it operates on is wrong for how I actually work. Obsidian thinks in notes. Files with backlinks. The graph is emergent decoration on top of unstructured text. Which means everything I capture is shaped like a document, even when it isn't. A meeting is a document. A decision is a document. A person is a document. A project status is a document. The structure lives in folder names and tag conventions I have to enforce by hand, every time, forever.
The thing I needed was the opposite. Typed entities with real fields. A Person has roles and a relationship to a Company. A Decision has a status, a date, an owner, and the options it foreclosed. A Project has a stage, a confidence rating, a horizon, a heat band derived from when I last touched it. None of this is novel database design. It's just refusing to pretend that everything in my head is the same shape.
Fibery gives me that. It is a structured database with documents living inside the entities, not the other way around. I can write a Decision node like a doc, but the Decision also knows it is a Decision, and views, queries, and my AI co-pilot can act on that. A Friday review can ask "what's overdue and important and not touched in two weeks" and get an answer, because importance and last-touched are fields, not vibes I have to remember to tag.
The deeper unlock is that this becomes addressable for AI. My instance of Claude reads from the graph by stable IDs. When I open a session it knows what a Stance is, what the Failure Log is, what's currently in the queue. I don't repaste context. The graph is the context. Obsidian markdown can be parsed, but the model has to infer schema every time. With typed entities the schema is the contract.
The cost is real. Fibery is heavier upfront. I had to think about what kinds of things I'm tracking before I could track them. Obsidian lets you start writing immediately, which is exactly why it ends up as a graveyard. The friction of declaring structure is the same friction that keeps it usable two years in.
So I'm not anti-Obsidian. I'm anti-flat-file as a model for thinking. If your work is mostly prose and you want fast capture with light retrieval, Obsidian is excellent. If your work is decisions, projects, people, money, and feedback loops that need to compound, you want types and fields, not folders and tags.
The second brain only earns the name when you can ask it questions it answers itself.
Ready to integrate AI into your business?
See how Model Context Protocol (MCP) can connect your AI assistant to all your business tools. Book a call with our team to discuss your specific needs.
Book a Call (opens in a new tab)