Make Your AI Read Your Brief Automatically

Once you have written the brief your AI should read every time, there is one more move that turns a good habit into an effortless one: stop pasting it.

Every serious AI tool now has a place to store standing context so it reads it automatically, before every conversation. You set it once, and from then on the tool already knows who you are and what you are working on. No more copy and paste. This is how to set that up in the three tools most people use.

A quick note before the steps: menus move around as these tools update. I am naming the feature in each tool, not exact click paths, so this still works after the next redesign. Search the settings for the feature name and you will find it.

If you use ChatGPT

This is the one most people should start with.

• Custom Instructions is your always-on brief. It is a settings panel where you tell ChatGPT, in plain text, who you are and how you want it to respond. Whatever you put there applies to every new chat. Paste the parts of your brief that are true across everything you do.
• Projects is for per-project context. A Project is a folder of chats with its own instructions and its own uploaded files. When one brief does not fit all your work, make a Project per thing you are building, and drop that project's brief into its instructions or as a file. ChatGPT reads it inside that Project automatically.

Rule of thumb: the general stuff goes in Custom Instructions, the project-specific stuff goes in a Project.

If you use Claude

• Projects on Claude work the same way. Each Project has a custom instructions box and a knowledge area where you can add files. Put your brief in the instructions, or upload it as a knowledge file, and Claude reads it for every chat in that Project.
• Claude Code (the version that lives in your terminal and builds software) reads a file named CLAUDE.md. If you drop a CLAUDE.md in your project folder, Claude Code reads it at the start of every session on its own. There is also a personal one that applies everywhere you work. If you are not writing code, you can skip this one.

If you use Cursor

Cursor is a code editor, so this is mostly for the building crowd.

• Rules are Cursor's version of the brief. You can set personal rules that apply everywhere, and project rules that live in the project itself. Cursor reads them automatically as it works. Put your brief in a project rule and you never paste context again.

The principle underneath all of them

The tools differ, but the idea is identical: there is a settings slot or a file that the tool always reads, and your brief belongs in it.

• One brief per project. A brief for everything is too broad to be useful. Give each real project its own.
• General context goes in the always-on slot (custom instructions, personal rules). Project context goes in the project (a Project, a CLAUDE.md, a project rule).
• When the AI keeps making the same mistake, do not re-explain it in chat. Add a line to the stored brief so it never happens again.

That is the whole upgrade. You wrote the brief once. Now let the tool carry it for you, and spend your energy on the work instead of the setup.