Image ingestion and conversation summaries

November 17, 2025

• Ingesting images is now supported and they will be visible when previewing a source in Unless. In the future, these will be used in AI answers as well. Tip: Alt-text and title tags will help the AI have some context to what the image is about.
• We made improvements to the system prompt to make it more strict against hallucinations.
• Filtering for escalated conversations is now possible.
• You can copy tasks to other use-cases.
• Conversation details view now includes a summary for each conversation.
• The disclaimer message is rendered as markdown so it is now possible to add links to it.
• Answer, search, and actions tab names are now automatically translated to the current conversation language in the AI component.
• When suggesting an answer from the team assistant, the last asked question will now be prefilled.
• Greetings and thank yous are excluded from unanswered questions.
• Insights now take time zones into account when reporting.