An AnalysisPlan tells the system how to:
You can mix or match these sub-plans. Each sub-plan has its own “messages” (the prompt content) and a “timeoutSeconds” (how long we wait before giving up).
Tip: We suggest using timeoutSeconds between 30 and 60 to give the AI enough time to respond.
call.analysis.summary
.enabled
: boolean (default true
).messages
: array of messages telling the model how to summarize.timeoutSeconds
: how long to wait. Generally set to 30–60 seconds.Example:
{
"summaryPlan": {
"enabled": true,
"messages": [
{
"role": "system",
"content": "Summarize this call succinctly."
},
{
"role": "user",
"content": "Here is the transcript:\\n\\n{{transcript}}\\n\\n"
}
],
"timeoutSeconds": 60
}
}
enabled
: boolean (default false
).schema
: the JSON Schema that describes the structure of the data you want.messages
: array of messages guiding the AI to produce valid JSON that fits the schema.timeoutSeconds
: how long to wait. We recommend 30–60 seconds.Schema Example: