Option to Manually Combine Summaries

[featurebase-for-github[bot]]

This issue is linked to our feedback platform. For feedback and updates, please visit this link.

[smian1]

Description

Sometimes, a single conversation results in multiple fragmented memories due to backend latency or system behavior. This can make it difficult to follow the full context of a discussion.

A solution would be to allow users to select multiple memories and manually combine them into a single entry. This would merge the transcripts and retrigger memory creation, producing a more cohesive summary.

Challenges

• Multiple memories from a single conversation can clutter the list and cause context fragmentation.
• No way to merge related memories that should belong to the same event.
• Loss of clarity when reviewing past conversations due to unnecessary separation of memory entries.

Proposed Enhancements

1. Manual Memory Combination

• Allow users to select 2 or more memories and flag them as being part of the same conversation.
• The app combines the transcripts and generates a new summary.

2. Memory Linking & Visibility

• Indicate that a memory is a combined entry (e.g., using an ℹ️ icon at the end of the memory headline).
• Link to the original memories so users can still reference them.
• Under default view, hide the original memories after merging.
• In "Show All" mode, display original memories in case users need to revert changes.

3. Status Updates for Merged Memories

• When two memories are combined:
  • Option 1: Keep original memories but flag them as "discarded."
  • Option 2: Move original memories to a secondary list while keeping the new combined memory as the primary view.

User Flow

1. Select Memories → Users choose two or more memories.
2. Combine → The system merges transcripts and regenerates a summary.
3. Update UI → Original memories are hidden (or moved to "discarded" status), and the new merged memory appears in the main view.

Benefits

✅ Provides better organization by reducing memory fragmentation.
✅ Ensures that conversations remain contextually intact.
✅ Gives users more control over how memories are structured.
✅ Allows for reversibility by keeping links to original memories.

[thestumonkey]

I've actually experienced the opposite problem. A conversation that goes on for so long that any summary is going to miss important points. I believe a good solution would be to enable multiple topics within a conversation.

A conversation can go on for hours after all, and currently I think it will only stop after a significant pause. Within a conversation, you can have multiple "threads" or topics. we would like to summarise these threads, and then have a conversation summary that encompasses the whole experierence.

If we do it that way, we may not need to join summaries as they will be bound within the same conversation container