Living on the agentic frontier
Point capable agents at your own data for long enough and you find out which bits of the hype are real.
rafeos
17Building a personal operating system in public
What a personal AI system costs
The real cost numbers for RafeOS, mid-build, and some decisions made along the way.
Agentic product development, up close
What building RafeOS with AI taught me about compression, discipline and judgement.
What it runs on
Infrastructure - a quick summary of the servers and devices running RafeOS
An agentic harness on swappable layers
A look at the agentic harness behind RafeOS. An account of what's working for me.
The second brain
How I've structured my markdown files for a shared substrate
AI-powered briefings
The layer that sits above all the modules and tells me which of them I should care about.
An attention layer, not a feed
RafeOS Reading module looks to give an AI and integration twist to the traditional feed reader.
Health, unified
How multiple health sources are combined into one and lessons from the agentic layer.
An agentic hybrid to-do list
What a shared AI-enabled and agent-friendly to-do list looks like. Lessons from RafeOS.
The messaging layer
Unified incoming comms and identifying what needs a reply.
One API, one MCP, one CLI
The one-ness at the heart of RafeOS is a shared API, MCP and CLI that make it surface agnostic.