Newly documented: how post-call outcomes are assigned
A new page explains how the call outcome is chosen after a call — and why outcome rules in the agent prompt have no effect.
- Two prompts, two jobs: the agent prompt runs the conversation; a separate AI
then writes the summary and the outcome. That AI does not see the agent prompt
— outcome rules belong in
configuration.postCallPrompt. - Copyable template for the post-call field, plus a checklist against the most common mistakes (voicemail landing as "Not interested" or "Callback requested").
- Doable via MCP: with
agents_get/agents_updateandcall_outcomes_list/call_outcomes_update.