AI agents
AI agent playbook
Use this guide when an AI agent sits between a user and the Activizor API. The agent owns the conversation; Activizor owns scheduling state.
Operating loop
- Send each relevant user scheduling message to chat.
- Inspect draft intents, active intents, and operations.
- Ask the user for the next missing detail when needed.
- Proceed to matching only after an intent becomes active.
Treat Activizor as the source of truth
Send relevant scheduling turns to Activizor and use the returned conversation state instead of reconstructing intent state yourself.
Clarify only blocking gaps
Ask for missing participant, date, or time details when Activizor keeps an intent in draft. Avoid extra questions once the intent is active.
Do not invent user commitments
Never create participants, time windows, or confirmations that were not provided by the user or returned by the API.
Match from active intents
Use active intents and their availability windows for matching. Draft intents are still conversational state, not matchable plans.