Using your own API credits

Make an API key in the Anthropic Console. Then in Nirvaana, open Settings, find “Your AI”, then “Use my own Anthropic API key”. Paste the key, set your monthly limit, and press “Check and save”.

• Nirvaana checks the key once with Anthropic. That check costs nothing.
• The key is kept locked, and never shown again: Settings shows only its last four characters, like “Your key ending x7Qa”. It goes only to Anthropic, never to your browser, the Mac app or any AI app you connect.
• Using your own API key isn't available for 13 to 17.

Your chats with Nirvaana, its quick answers, the Team chat answers you ask @Nirvaana for, your automations and scheduled asks, drafting an automation, reading the inbox you asked it to keep checking, and your study lessons and course facts.

Nirvaana's own safety checks and its background work stay on Nirvaana: the standup, the daily tidy and the nightly check it runs when you first open the day, and what it says after your cloud computer finishes a task. So does anything that isn't a model answer, like your cloud computer's time, and those still keep to your plan's usual limits.

Settings shows how much of your limit this month has used. What runs on your key doesn't count toward your plan's own use.

If Anthropic doesn't accept your key, your credits run out, or this month reaches your limit:

• Out of the box, Nirvaana stops and uses nothing of its own. In your chats, scheduled asks, automations and lessons it tells you why, and so does a mail it didn't read. In Team chat, @Nirvaana just doesn't answer.
• If you turned on “If my key stops working or reaches my limit, use Nirvaana's instead”, it carries on with Nirvaana's own, within your plan's usual limits.

Settings says what went wrong. Paste the key again, or a new one, once it's fixed.

Press “Remove” under your key in Settings. The key is deleted from Nirvaana at once, and its answers run on Nirvaana's own again. Deleting your account deletes it too.