Gyaanmitra AI
Privacy & Data
What stays on your machine, what is sent to the model provider you chose, and how to review or remove stored history.
The short version
Gyaanmitra AI is bring-your-own-key. Chat requests go from the desktop app to the provider you connected. Prompt content is not routed through a hosted Gyaanmitra model pool.
Read the upstream model's data-use terms for the provider you pick.
Stays on your machine
In the desktop app, conversation history, your rules, and your API keys are stored locally in your Gyaanmitra profile. See Portable Mode & Uninstalling for the exact paths.
Sent only to your chosen provider
The context in a request (the files it read, your message) goes to the model provider you set up, under that provider's own data-handling terms.
Your model provider's own privacy policy still applies
Gyaanmitra AI does not add its own data-retention layer on top of your provider, and it cannot override theirs either. Read the terms of the provider you connect. Retention and training-use rules differ between providers, and sometimes between models from the same provider.
What's included in a request
Only what is relevant to the conversation you are in: the message you send, any file or context you mentioned or that is currently open, and, if you are continuing a conversation, the recent turns of that same conversation. Gyaanmitra AI does not scan your whole workspace or send your whole codebase on every message. See Tools, Rules & Context for how context is built.
Managing stored history
Review past conversations
Open Chat History from the clock icon in the Gyaanmitra AI panel to see and reopen previous conversations. History stays in the desktop app.
Delete a conversation
Remove one conversation from Chat History, or clear everything at once from Gyaanmitra AI Settings → Features.
Remove an API key
Delete a stored provider key any time from Gyaanmitra AI Settings → Providers in the desktop app. Further requests can no longer use that key.
Workspace Trust and Gyaanmitra AI
If a workspace has not been granted Workspace Trust, Gyaanmitra AI's ability to run terminal commands or apply edits is limited the same way other workspace automation is. That keeps an untrusted folder from steering what runs on your machine.