Building a Life OS with Obsidian and AI
Why a Life OS
Most productivity systems fail because they demand too much maintenance. You spend more time organizing your tasks than actually doing them. I wanted something different: a system that mostly manages itself, nudges me when something needs attention, and stays out of my way the rest of the time.
The result is what I call my Life OS — a combination of Obsidian as the knowledge base, AI agents for automation, and a set of routines that keep everything running smoothly.
The Obsidian Foundation
Obsidian serves as the single source of truth. Everything lives here: project notes, daily journals, meeting records, reading highlights, and personal goals. The key design principle is that information should be captured with minimal friction and organized automatically wherever possible.
Folder Structure
I keep the structure deliberately flat. A handful of top-level folders handle the main areas, and tags plus links do the heavy lifting for cross-referencing.
/
├── Projects/ # Active project notes
├── Areas/ # Ongoing areas of responsibility
├── Resources/ # Reference material
├── Journal/ # Daily notes
├── Templates/ # Note templates
└── Archive/ # Completed projects
Templates and Automation
Every new note starts from a template. Daily notes auto-populate with my calendar events, pending tasks, and any items flagged for review. This removes the friction of starting each day — I open the daily note and everything I need to know is already there.
The AI Layer
This is where it gets interesting. I have a set of AI agents that run on schedules and handle different aspects of the system.
The Guardian Agent
Runs every few hours and checks the overall health of the system. It looks for orphaned notes, stale projects, and anything that might be falling through the cracks. When it finds something, it sends me a notification with a suggested action.
The Worker Agent
Handles the actual grunt work: summarizing meeting notes, extracting action items from journal entries, updating project statuses, and generating weekly reviews. It reads context from the workspace and produces structured outputs that feed back into Obsidian.
The best automation is the kind you forget exists. It just works in the background, making your system a little better every day.
Lessons from a Year of Use
The biggest lesson is that simplicity beats sophistication every time. Early versions of the system were over-engineered with complex workflows and too many automation triggers. Stripping it back to the essentials — capture, organize, review — made it sustainable.
What Makes It Stick
Three things keep me using this system daily. First, the capture friction is near zero. Second, the AI agents handle the boring parts I would otherwise skip. Third, the weekly review habit closes the loop and prevents drift. Together, these create a system that actually gets better over time instead of slowly decaying like every other productivity system I have tried.