Overview
AI prompts power LLM-assisted steps in automations and suggest-reply in Inbox. You can use platform-managed models or bring your own keys (OpenAI, Azure OpenAI, or Gemini-style providers). Keys are envelope-encrypted and never returned in plaintext. Outputs pass moderation before send; hourly token budgets protect runaway spend.
Where to configure
Open Dashboard → AI to manage prompts and keys, then reference prompts from automation actions. Usage and cost signals appear in Analytics when AI calls run. Plan entitlements gate AI features and reply limits.
Was this article helpful?
Your feedback helps us improve our documentation.