The Ask node speaks a question and captures what the caller says, transcribed to text by speech recognition (STT). It re-asks if the caller is silent, accepts a list of expected words to improve recognition, and saves the answer under a question key. Later nodes can use the answer as {{answer.<key>}}, and it appears in call logs and the status callback.

Settings#

Field	Required	Default	Notes
Question key	Yes		Names the answer. Lowercase letters, digits and `_`, starting with a letter, e.g. `location`. Must be unique in the flow
Question	Yes		Spoken question. Supports `{{var.name}}`. TTS preview available
Language	No	Flow language	Default, English (India), English (US) or Hindi. Used to speak the question and to transcribe the answer
Wait for an answer (seconds)	No	8	1 to 30
Pause that ends the answer (seconds)	No	2	1 to 5. Raise it for open questions, lower it for one-word answers
Re-ask if silent	No	1	0 to 2 extra attempts before following No answer
Re-ask prompt	No		Spoken instead of the question when re-asking
Expected words (optional)	No		Comma-separated names or terms callers are likely to say, e.g. `Delhi, Lucknow, Sales` (up to 50)

Exits#

Exit	When
Answered	A spoken answer was captured
No answer	The caller stayed silent after all re-asks

If the Answered exit isn't connected, the call hangs up after the question (the builder warns you).

Where answers appear#

{{answer.<question key>}} in later HTTP Request and Email nodes.
The Answers section of the call in Call Logs.
The answers array of the status callback: question_key, question, transcript, confidence, status, attempt.

Example#

A Hindi lead-qualification step:

Field	Value
Question key	`city`
Question	Aap kis shehar se baat kar rahe hain?
Language	Hindi
Expected words	Delhi, Mumbai, Lucknow, Jaipur

Then an HTTP Request sends {"city": "{{answer.city}}"} to your CRM.