What is a personal AI operating system?
A personal AI operating system is a set of files and routines you build around an AI assistant. Keep instructions and project notes in a folder, then configure the assistant to read them when it starts work. You can use those notes to prepare a morning briefing or revisit a decision without explaining the whole project again.
The point is to keep context somewhere you can review and update. Saved files make it easier to carry information between sessions, but they need maintenance. Correct stale notes, remove details you no longer want it to use, and check which files it actually read.
You can start by describing what you want to Claude Code or Codex. The first project can be a single markdown file. Connected tools and scheduled runs take more setup, so add them when the simple version is working. I’m a designer, and this is how I built the briefing I use in the mornings.
The five layers
Instructions it can read
Write down what you want help with and how you like to work. Configure your assistant to read those instructions when a session starts. If you need afternoons free or prefer a short answer, put that in the file. Check what it follows and adjust the parts it misses.
Notes you can carry forward
Ask the assistant to save a decision or useful detail in memory.md. At the next session, ask it to read those notes before starting work. Review the file occasionally, especially when a project or preference changes.
The daily briefing
Start with a note about your day and ask for a short briefing. Once the format helps, connect your calendar and add a schedule. The assistant can then prepare the file before you sit down to work, using the sources you have connected.
Tasks routed by energy
Tell it when you tend to have energy for difficult work. Ask it to plan around that and your fixed commitments. After a rough night or an unexpected meeting, update the plan with what you can manage today.
A coaching layer
Ask it to look for a pattern you care about, such as a task that keeps moving to next week. Have it point to the notes behind its observation so you can check whether it has understood. You decide what to do with the suggestion.
Why keep your context in files?
A plain-text file is easy to inspect. You can change a line, keep a backup, or bring it to another assistant. The new tool may need different instructions and connections, but you have the notes to start from. You are also responsible for keeping them accurate.
Keeping files on your computer does not mean everything stays there. A cloud AI provider still processes the information you send it. Choose which files and tools the assistant can access, and check what a connection shares before adding it.
A briefing is also a manageable first project. You know what happened today, so you can tell whether the answer is useful or wrong. That practice helps when you move on to building a website or an app.
Start with one file
Open a project folder in Claude Code or Codex and create a file called about-me.md. Write a short paragraph about your work and what you want help with. Ask the assistant to read that file and suggest a first task. Check its answer against what you wrote, then correct anything it misunderstood. The free personal agent guide walks through the next steps. Recorded course lessons use Claude Code; setup and commands need adapting in Codex.