Skip to main content
The AI assistant is available from the lower-right corner of the signed-in console. Ask how to use OmniRelay features, keys, models, usage pages or API integrations. It is a console help feature, not a public inference endpoint or a way to run models through your account.

​
What it can answer

The assistant uses reviewed public OmniRelay guidance to answer product how-to questions and common troubleshooting questions. It is designed to decline unrelated questions and requests for private or sensitive account information. It does not query your workspace, balance, keys, requests, customer records or permissions. Use the relevant console page for information about your own account, and contact support when you need help with a specific record. The answer is general guidance. It does not change settings, create keys, run API calls or perform actions in your account. Check your own console and /v1/models response for current access, prices and limits.

​
Data and credentials

The assistant sends your latest question and up to three recent messages from the same conversation, together with selected public product guidance, to the model provider that OmniRelay uses for the assistant. If the platform assistant is not configured, the assistant can instead use your selected OmniRelay inference key and bill it as a normal model call. OmniRelay does not save assistant conversations in its database; the temporary conversation is held in your browser and clears when the page reloads, the account changes or you start a new chat. Do not send passwords, API keys, login codes, payment details or private customer information. The assistant cannot verify account-specific facts. For how request data is handled, see Data handling.

​
Language and availability

The assistant is available to signed-in users when OmniRelay has enabled it. It responds in the language of your question where supported. If configuration is unavailable, use the contextual help and developer documentation instead.