Customization

Conversation Types

Differentiate your calls to tailor prompts, automations, and analysis frameworks.
Definition: Conversation types let you classify your calls by business use so the right context, prompts, and automations are applied.

Feature video

No video is available at this time.
Who is this for?Required permissions and licenses
• Teams that cover several types of exchanges within the same Leexi workspace.
• Business administrators who want to differentiate AI outputs based on context.
• Available on AI Meeting, Business, and Enterprise plans. Administrator role required for management.
• The number of prompts per type depends on the plan:
AI Meeting: 1 generic summary + 2 custom summaries
Business: 1 generic summary + 4 custom summaries
Enterprise: 1 generic summary + 8 custom summaries
• Conversation types work together with prompts, AI context, scorecards, and CRM Autofill.

How to use it

Identify your main use cases

List the families of conversations that deserve differentiated handling.

Create the useful types

Go to Settings > Customization settings > Conversation Types and click "Add". Default types already exist (internal meetings, generic meetings). Fill in the type's information, including the definition for the AI and the title keywords for automatic recognition.

Associate each type with the right AI context, prompts, CRM fields, or scorecards.

Test on real calls

Check that the outputs produced match the expected context.

Going further

  • Conversation types serve as the backbone for many Leexi customizations.
  • A simple taxonomy is often more effective than too many types.
  • They are especially useful for separating sales, support, recruitment, or management journeys.
  • Automatic recognition relies on two settings: the definition for the AI (which analyzes the transcript) and the title keywords (based on the meeting's name in the calendar).
  • Advanced options let you refine recognition further: call type (video, phone, upload), internal/external, direction (inbound/outbound), tags, deal status, and duration.
  • If the AI applies the wrong type, you can always correct it manually from the dropdown menu at the top of the call page.

Frequently asked questions

Things to watch out for

  • Too many types make maintenance difficult.
  • A type without useful context or a prompt brings little value.
  • Test automations on a small scope before rolling out broadly.

Previous and next pages

Copyright © 2026 Leexi