context-use: turn your data exports into portable AI memory

All your online interactions (your prompts, your searches, your stories, your messages) are stored in a handful of platforms and available to you via a simple data export.

context-use is the tool to turn any data export into portable context so that your favorite AI agents can act less like boring chatbots and more like friends who truly understand you.

With context-use, you can:

• process any data export (Instagram stories, YouTube videos, Google emails, etc.) into AI memory

• port your memories between ChatGPT, Claude, Gemini and any other LLM

• run locally using your favorite LLM provider so nobody else sees your data

And it’s fully open-source!

Recently, Anthropic launched Claude Import Memories to enable people to switch from ChatGPT to Claude while carrying their memories with them.

The memories generated by context-use are much richer than the ones generated by ChatGPT when requesting them with a prompt (like Claude Import Memories suggests) because context-use processes the full data export, not just the cherry picked memories.

Getting started is super easy:

context-use is packaged as a CLI, install it with:

pip install context-use
# or
uv tool install context-use

Export a copy of all your ChatGPT data to your computer by clicking on “Export data“ of the Data Control page. For more information follow this guide.

Once your ChatGPT data export is ready, you will receive an email to download the data export as a zip archive. Store it on your local machine.

To easily generate memories from your zipped data export, run:

context-use pipeline —-quick <your_zipped_data_export>

The —-quick flag only generates memories from your last 30 days of data. The memories generated will be exported as a markdown file, ready for you to use.

This is just the quick start command, to process the full data export remove the —-quick flag. For more information, refer to the README.

Once your memories have been generated you can simply copy/paste them into Claude’s Capabilities page.

After Claude has processed the memories you generated with context-use it knows who you are, what you are building and every other conversation you had with ChatGPT.

context-use allows you to turn your data archive into AI memories but the process of doing so is quite cumbersome and lacks critical features:

• no programmatic transfers: manually exporting your data from a source and uploading the memories to a destination is a hustle

• no continuous updates: your Digital Self evolves every day as you interact in the digital world but your data export is only a static snapshot

• no semantic scopes: each agent should only have access to what it needs and only you should decide who has access to what

At Fabric we spent years working with Big Tech companies and regulators to make it easy and secure for you to gather all your data, curate your context, and port it to all your favorite AIs.

We will soon be launching a hosted version of context-use which enables you to programmatically transfer your data, keep your Digital Self in sync and grant scoped access to your memories. Sign up here if you want to be among the first to try it out!

Bringing memories from ChatGPT to Claude is cool but context-use enables you to go much further than that.

context-use helps you create your Digital Self from all your online interactions. ChatGPT knows one aspect of who you are, Instagram knows a different side, and Google yet another, but none of them knows the real you.

With the Claude Import Memories announcement, the battle for your context has started. The AI labs are fighting for your context as if it were a scarce resource.

context-use turns the context battlefield into common ground. When your context is portable, the incentives change. Instead of a zero-sum race for control, companies are forced to focus on what actually matters: adding value to you.

For some inspiration on what’s possible when AI agents can access your context: