So, I finally decided to build a KMS (Knowledge Management System). I’ve been sitting on the idea for a while, but this week I actually sat down to plan it out properly.
And immediately, reality hit.
The deeper I looked into the architecture, the more I realized my initial “how hard can it be?” optimism was… misplaced. There are some serious technical hurdles here, especially with the Local-First approach.
The Mobile Problem
I want this thing to be fast and local. But treating a desktop and a phone as equals in a peer-to-peer system is tricky. Phones burn battery, have less memory, and aggressive OS restrictions.
I can’t just port the desktop app to mobile and call it a day. I have to accept that the mobile experience will be a “consumer” first, and a “creator” second. Syncing thousands of Markdown files and running vector search on an iPhone without draining the battery in an hour is a non-trivial problem.
The Business Model (Or: How to Eat)
I’m an indie hacker now, which means I have to think about sustainability from Day 1. I’m torn between “information wants to be free” and “servers/APIs cost money.”
Here is where I’m landing:
The Free Tier (The Essentials) Actual knowledge management should be accessible.
- File Management: It’s your data.
- Basic Search: Finding things by name or tag.
- The Editor: Writing without friction.
These will always be free. I’m not going to hold your notes hostage.
The Pro Tier (The Heavy Lifting) This is where the costs (computational or otherwise) come in.
- Advanced Full-Text Search: If it needs an indexer running in the background.
- Similarity Search: “Show me related notes” (Vector embeddings).
- AI Features: Auto-tagging, summarization, connecting the dots.
Still Figuring It Out
I haven’t written the first line of core logic yet, and I’m already negotiating features with myself. But that’s part of the game.
I’ll keep you posted as I start breaking ground.
Note: I’m a solo developer based in Korea. To share my journey with a wider audience, I used AI to help translate my thoughts into English. If any phrasing feels a bit “too AI” or unnatural, please bear with me.