stash.

AI memory on your device vs in the cloud

Every assistant is adding a memory. It is worth knowing where that memory lives and who can reach it.

What built-in memory is good at

It is convenient. The assistant remembers your preferences and some facts about you without any setup, and it uses them in later answers. For light personalisation that is often enough.

Where it falls short

• It is per product. What you taught one assistant does not follow you to another.
• It lives on their servers. You are trusting the provider with a running record of what you think about.
• It is a summary, not an archive. You cannot reliably pull up the exact conversation where you made a decision.
• It does not include what you read. The article that started the chat is not in it.

What a local memory changes

A memory kept on your own device is yours to search, export and delete. It spans every assistant you use, and it can hold the pages you read alongside the chats. Because it never leaves your machine, there is nothing to breach and nothing to sell.

Stash is built this way. Saving, search and resuming a chat are free. Stash Pro adds a small model that runs inside the extension so you can ask a question and get an answer assembled from your own saved material, with every line citing where it came from. It is a one time purchase with no subscription.

Use both

This is not either or. Let the assistant's memory handle preferences. Keep the work itself, the decisions, drafts and research, somewhere you control.

More guides: How to search your old ChatGPT conversations · How to save your ChatGPT, Claude and Gemini chats · How to continue an AI chat in a new thread