Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.talkturo.ai/llms.txt

Use this file to discover all available pages before exploring further.

After every call, Talkturo’s AI info extractor reads the full conversation transcript and pulls out any relevant facts it finds — email addresses, company names, stated preferences, product interests, and more. It then writes that data directly into the CRM contact record, either populating existing custom fields or creating new ones on the fly. You get an enriched contact record without any manual data entry.

​
How it works

1

The call ends

When an AI assistant finishes a call, Talkturo saves the full call transcript.
2

The AI reads the transcript

The info extractor analyzes the conversation and identifies facts the caller shared — for example, their email address, the company they work for, a product they mentioned, or a preference they expressed.
3

Fields are matched or created

For each piece of data found, the extractor checks whether a matching custom field already exists on the contact record:
  • If a matching field exists, it writes the value into that field.
  • If no matching field exists and Auto-create missing custom fields is enabled, it creates a new custom field and populates it.
4

The contact record is updated

The extracted values appear in the contact’s custom fields. Open the contact record in your CRM to review what was captured.
Info extraction runs automatically at the end of every call — you do not need to trigger it manually. There is no delay; the data appears in the contact record shortly after the call completes.

​
What gets extracted

The extractor captures any factual information a caller shares during the conversation. Common examples include:

Contact details

Email addresses, phone numbers, and names mentioned during the call.

Company information

The organization the caller represents, their role, or their industry.

Preferences and interests

Products or services the caller expressed interest in, timelines, and stated needs.

Any other mentioned data

Any fact the caller shares that the AI recognizes as structured, storable information.
The extractor does not have a fixed list of fields it looks for — it reads the full conversation and decides what is worth saving based on context.

​
Auto-create missing custom fields

When the extractor finds a fact that does not match any existing custom field, it can create a new field automatically. New fields are always created as text type. You can change the field type afterward in your custom field settings if you need a number, date, or select field instead. To disable auto-creation (so extraction only writes to fields you have already defined), turn off Auto-create missing custom fields in your assistant’s Functions tab.
Start with auto-creation enabled. After a few calls, review the fields that were created and refine their types and labels. Then you can disable auto-creation once your schema is stable.

​
Enable info extraction on an assistant

Info extraction is a per-assistant setting. You configure it in the Functions tab of each assistant.
1

Open your assistant

Go to Assistants in the sidebar and select the assistant you want to configure.
2

Open the Functions tab

Click the Functions tab in the assistant editor.
3

Enable Info extractor

Find the Info extractor section and toggle it on.
4

Choose extraction targets

Check Save to Contact to write extracted data to the contact’s custom fields. You can also enable Auto-create missing custom fields to let the extractor add new fields when it finds data with no existing match.
5

Save your assistant

Save the assistant. All future calls handled by this assistant will run info extraction automatically.

​
View extracted data

After a call completes, open the contact record in the Contacts page. The custom fields section shows all values the extractor wrote, alongside any fields you had populated previously. Each field displays its current value — you can edit any value manually if the extraction result needs correction.

​
Tips for better extraction

Tell your assistant what data to collect. Add a line like the following to your assistant’s system prompt:
“During the conversation, gather the caller’s email address, their company name, and what product they are most interested in.”
When the AI knows what to listen for, it captures data more consistently and accurately.
Name your custom fields clearly. The extractor matches extracted facts to existing fields by label. A field named Product Interest is more likely to receive a correct match than one named Field 3.
Info extraction is a plan feature. If your current plan does not include the Info Extractor, the toggle will not appear in the Functions tab. Check your plan limits under Billing to confirm access.