Skip to main content
Questions are what the agent proactively asks the caller. Data extraction is the structured information the agent pulls out of the conversation so you get clean, usable records after each call.

​
Questions

1

Open the Questions tab

Go to /agents, open an agent, then select Questions (/agents/:id/questions).
2

Add questions

Click Add question and type each question you want the agent to ask. You can add up to 20.
Click the generate button (the sparkles icon) to suggest a starter set of questions you can edit or trim.
3

Reorder and remove

Use the trash icon to remove a question. The order you list them in is the order the agent will tend to ask.

​
Data extraction

1

Open the Data Extraction tab

Select Data Extraction (/agents/:id/data-extraction).
2

Define fields to capture

Click Add Field and name each piece of information you want extracted — for example, budget, preferred_time, or company_size. Use the generate button (the sparkles icon) to suggest fields.
After calls, the values for these fields are captured from the transcript and available with the call record.

​
Troubleshooting

Both features need the agent’s name and language set. Configure those on the Overview tab first.
You can have at most 20 questions, and you can’t add a new one while an existing question is left blank. Fill or remove empty rows first.
The agent can only extract what was actually discussed. Keep field names clear and make sure your questions prompt the caller for that information.