Skip to main content
Presentation memory keeps useful context connected to one deck. It helps the assistant remember the original brief, source documents, outline, previous edits, and earlier messages in the same conversation.

​
What this enables

  • Follow-up requests such as “use the same tone on the next slide”
  • References to facts from an uploaded document
  • More consistent edits across a longer conversation
  • Better awareness of earlier outline and slide changes

​
Where it is stored

The default setup uses Mem0 OSS with local Qdrant and SQLite storage under your persistent application data. You can keep it enabled, configure its models and storage paths, or disable it.
Persist /app_data. If the container is replaced without that volume, presentations, uploaded assets, and local memory can be lost.
Memory is supporting context, not the visible deck itself. Always review important content after an AI-assisted edit.

Configure memory and storage

Review Mem0 defaults, storage paths, embedding settings, document parsing, and backup guidance.