Most businesses win not because they have more features, but because they communicate a clearer philosophy—a picture of the future that some people want to live inside.
In older eras, religion acted as the gravity of communities. Today, we're still pulled by gravity, just different kinds: interests, crafts, identities, missions. People want to become good at something. They want to enjoy something deeply. They want to find tools that make their journey feel coherent.
A startup's job is to make a specific journey possible and to make it feel inevitable.
Notes Are Not Productivity. They're Mind Extension.
Note-taking is one of the most underestimated categories in software because it looks simple. But notes aren't documents or content. They're closer to memory, and memory is closer to identity. A note-taking system is an extension of your mind: a place where thoughts land, take shape, connect, and return later as insight.
The mind operates in modes. There's private thinking—personal thoughts, half-formed ideas, internal reflections. There's work mode—decisions, planning, collaboration, outcomes. There's publishing mode—writing for other people, making ideas legible, shipping something public.
Good note-taking tools don't force those modes into a single abstraction. They let them coexist.
This is why many serious note-takers end up with multiple tools:
- Obsidian for long-term knowledge and personal/work thinking
- Logseq for daily bullets and chronological capture
- A dedicated folder for meeting notes, recordings, and attachments
These aren't competing products. They're compatible because they share a common substrate: Markdown files on disk.
Files Are a Human Interface, Not a Technical Detail
File-based systems are a cognitive model, not a relic. Humans have organized information spatially for centuries: drawers, cabinets, boxes, folders. We remember where something is. Location is recall. Structure is memory.
This is why the terminal still feels natural. ls, pwd, grep—these commands mirror how people already understand information. There's a place, there's a name, there's content, there's a way to find it. Computers didn't invent this model. They adopted it.
When we talk about "local-first" and "file-based," we're discussing something beyond implementation: trust. A note-taking system is a memory prosthetic. Memory needs to feel owned, reliable, and genuinely yours. When notes live as files, they feel like part of you. When notes live as opaque blobs behind an API, they feel like someone else's product.
Why AI Works So Well in Coding (and Why Writing Is Following)
Coding agents feel effective right now partly because models are capable, but also because code is file-based. Software lives as artifacts—foo.tsx, bar.py, main.rs, README.md—not rows in a database or JSON stuffed inside Postgres fields. Real files with real names, clear structure, readable content.
AI agents can traverse repositories, read context, understand conventions, search across files, and generate changes that fit the existing structure. The agent sees the artifact, diffs it, rewrites it, reasons about it.
The best coding tools work because they're not inventing a new abstraction. They ride the filesystem—the same abstraction humans have used to manage complexity for decades.
Writing is following the same pattern. People draft blog posts and essays inside coding environments. It seems odd on the surface, but writing and coding are closer than we admit. In both cases you're composing, shaping intent into an artifact, embedding values into structure, turning a vague thought into something executable—either by machines (code) or by humans (writing).
"Agents for writing" are happening inside code tools because those tools are file-native. They treat writing as a first-class artifact, not as a database record.
The Endgame: Less Folder Worship, More Meaning
The future probably isn't "more folders." Folders are useful as a spatial index for recall, but they're a human workaround.
In the AI era, titles matter less than content. Location matters less than meaning. The real organizing principle will shift toward content semantics, metadata, search, retrieval, and links between ideas. The filesystem remains the substrate. The intelligence layer sits on top.
Where Char Fits
In the note-taking ecosystem, Obsidian communicates a clearer picture of the future than most. It's an incredible interface on top of a file-based knowledge base, and it has influenced how an entire generation thinks about notes. I've subscribed for years because of what it represents, not because of one feature.
We don't want to replace Obsidian. We want to be part of the same movement: tools that respect files, respect ownership, and respect the mind.
Char is built to become the best interface for meeting notes and a connector across the full meeting workflow. Before the meeting: context, agenda, prep. During: capture, audio, real-time structure. After: summaries, decisions, action items, follow-ups. Eventually, a command center where you can direct AI instead of being replaced by it.
Because Char is file-based and local-first, it can sit alongside your existing vault. It can read from your knowledge base and write back into it. It treats meetings as a specialized region of your brain—deeply integrated, but not controlling everything else.
That's the direction that matters: building a tool that fits into the cortex, not another walled garden.
Closing
This isn't nostalgia for old-school files. It's a belief about the future: if AI is going to help humans think, it needs access to artifacts humans can own, inspect, move, and understand. Markdown, plain text, files.
The filesystem is the cortex. In the AI era, it becomes even more important, not less.
Char is our contribution to that future: a meeting-native interface inside a file-native world.
