FROGED | Summarise conversation

Help

Summarise conversation

In the FROGED inbox, artificial intelligence features have been added. One of them is to summarise the content of the conversation in several lines that summarise the key points. All agents in the workspace have access to this functionality if they have it configured in their settings and permissions.

To summarise a conversation in the inbox you must click on the button at the top ‘AI summary’.

When this button is clicked, the summary is generated in a private note within the conversation. This message is only visible to the agents working from FROGED, not to the client.

The parameters contained in the summary are:

Main language of the conversation
Main customer issues or requests
Key actions taken or solutions provided
Pending items or follow-ups needed
Customer sentiment
Agent sentiment

To generate a conversation summary in the inbox correctly it must meet the following requirements:

At least five conversation elements of types supported by the summary are required. Not every history event counts; internal notes are excluded by that filter.
Audio, attachment and transcript availability depends on the specific flow. A file reference does not guarantee that the summary reads its entire contents.
Do not assume this summary includes a complete call transcript. Check the voice channel's summary and transcription features separately.

Introduction to FROGED omnichannel inbox

Assign conversations to agents and teams

Close conversation

AI agent configuration

AI agents are currently configured with the FROGED support team. Share your objective, channels and examples of conversations you would like to review. Do not include credentials or unnecessary customer data. Conversation summaries and AI agents are separate features.

Summary requirements and limitations

Generating a summary requires the plan's AI capability and Generative AI permission. The result is added as a conversation note; review it before using it as the basis for a reply or decision.

Processing requires at least five conversation elements of supported types. Not every history event counts: internal notes, for example, are excluded by the filter this summary uses.

For a file or document, the summary may receive its reference or description in the conversation; do not promise that it automatically reads the entire attachment. Do not claim this method includes a complete call transcript without checking the specific voice flow.

If “conversation too short” or “no content” appears, review the available messages. For another error, keep its text and the conversation reference for support. Do not repeatedly generate notes without checking whether a previous request already added the summary.

A summary helps review context, but does not replace original messages or guarantee that AI interpreted every detail correctly.

Was this article helpful?

How can we help you?

Related articles

AI agent configuration

Summary requirements and limitations