The Cluster Map turns your library into a 2D map of topic clusters and similarity neighbourhoods. Use it to see at a glance what you actually have, where the gaps are, and which papers are duplicates of each other in everything but title.
What you get
- A cluster view computed locally from your saved articles.
- Adjustable clustering depth — coarse for an overview, fine for nuance.
- Per-cluster summaries and a quick comparative card for member papers.
- Drill-down into a cluster opens the matching papers in the library.
Main buttons inside Cluster Map
| Button | What it does | | --- | --- | | Cluster | Rebuilds the map with the current parameters | | Suggest K | Helps estimate a useful cluster count | | Quick Compare | Compares representative papers from the selected cluster | | Filter to Cluster | Narrows the visible library items to one cluster | | Summarize | Generates a summary of the selected cluster | | Research Brief | Produces a more structured cluster-level brief | | To Writer | Sends the current cluster narrative into Writer | | Compare Clusters | Puts two clusters side by side |
Workflow
1. Open the Cluster Map dock from the Help/Landing capability cards or the View menu. 2. Pick a depth — the map re-clusters live as you change it. 3. Click a cluster to see its papers and a quick comparative summary. 4. Use the comparison to merge, prune, or extend a research thread.
Notes
Clustering runs locally on top of the same embeddings the AI assistant uses. The result is private to your library and never leaves the machine unless you export it.