Post-call data extraction

What is post-call extraction?

Instead of reading transcripts to find the caller's name or the reason they called, you define the fields you want once, and every call's transcript is analyzed automatically after it ends. The values land on the call's page in the dashboard as structured data - and in your webhook payloads if you use them.

How do I set up the fields?

1. Open your agent's Tools tab. Under "Data to Collect", add a field: a name (for example patient_name) and a description of what it is or how to ask for it ("Ask for the caller's full name").
2. Add as many fields as you need - name, callback number, appointment date, reason for calling, budget - and save.
3. Mentioning the same information needs in the prompt helps too: the agent asks during the call, and extraction picks the answers out of the transcript afterwards.

Where do the extracted values show up?

• On each call's page in the dashboard, alongside the transcript, summary, and sentiment.
• In the payload of your organization's outgoing webhooks, under extracted fields - ready for your CRM or spreadsheet automation.
• In after-call tool payloads as collected values (see: Custom tools and webhooks).

What if a value wasn't mentioned on the call?

Fields the caller never provided simply come back empty - nothing is invented. If a field matters, make sure the prompt tells the agent to ask for it explicitly.