NotaeLibrary
Turn your study room's World entities, bookshelf documents, and deep research into a beautifully cross-referenced, AI-generated encyclopedia — with inline citations back to your team's evidence.
Create and manage your knowledge Worlds.
How Worlds work — entities, types, and settings.
Visualising and rearranging entity relationships.
Notaepedia is an AI-generated encyclopediafor your study room. Every person, place, event, piece of evidence, and context entity in your room's World gets its own article — written by Sage, citing your team's actual research.
Think of it as Wikipedia, but for your investigation. The articles are cross-referenced with links to other entities, and every claim is backed by inline citations pointing to the specific bookshelf document, uploaded notae, or deep research report that provides the evidence.
When you generate (or regenerate) an article, Sage combines three sources of information:
Creating a Notaepedia follows a natural progression. Each step builds on the last.
A World is a standalone knowledge base — a collection of entities with typed relationships between them. Multiple study rooms can share the same World, and a single room can link to multiple Worlds.
Once your World is created, add entities via three methods:
Paste a Wikipedia article URL. AI reads the article section by section and extracts structured entities — people, places, timeline events, evidence, and context — along with the relations between them. After extraction, review and approve which entities to add to your World. The system automatically detects and consolidates duplicates within the extraction.
Click “New Entity” on your World's entities page. Enter a name, choose an entity type (Character/Person, Place, Timeline Event, Object, or Context), add a description, aliases, and metadata. As you type the name, the system checks for existing similar entities to prevent duplicates.
Any document on your room's bookshelf — a deep research report, a web article, a YouTube transcript, an uploaded PDF, or a book — can be promoted to a World entity. The document itself stays on the bookshelf as source evidence; the promoted entity becomes a node in your World graph.
| Type | Use For | Examples |
|---|---|---|
| CHARACTER | People, suspects, witnesses, investigators | Paul Fronczak, Detective Miller, Dora Fronczak |
| PLACE | Locations, buildings, cities, regions | Michael Reese Hospital, Chicago, the alleyway |
| TIMELINE | Dated events, milestones, key moments | April 26, 1964 — kidnapping discovered |
| OBJECT | Physical evidence, items, documents | Birth certificate, hospital bracelet, DNA swab |
| CONTEXT | Theories, background, legal frameworks | The Imposter Theory, hospital security 1960s |
| WORK | Books, articles, films being analysed | The Fronczak Case (book), 20/20 episode |
After creating entities, link them together with typed relationships. In your World, go to the Relations tab or use the Entity Graph to draw connections visually.
Relations have a label(free-text like “was last seen at,” “is married to,” “investigated by”), an optional source sentence (the exact passage from your evidence), and a confidence level (1–5). These relations appear in every Notaepedia article for that entity and create the cross-reference links between articles.
The bookshelf is where your team's evidence lives. Every document added to the bookshelf gets vectorized — broken into chunks and embedded into a semantic search index. When Sage writes a Notaepedia article, it retrieves the most relevant chunks and cites them.
This is the most powerful path. Here's how it works:
Each bookshelf item has a confidence rating (1–5). When setting up your Notaepedia, you can set a minimum confidence threshold — only sources rated at or above that level will be used as citations. This lets you distinguish verified evidence (confidence 4–5) from unverified leads (confidence 1–2).
With your World populated and bookshelf stocked, you're ready to generate articles.
Room owners and moderators can add moderator notes to any article — corrections, additional context, or warnings about disputed facts. These notes appear at the top of the article and are visible to all readers.
Once your Notaepedia is complete, you can export the entire collection as an EPUB ebook — complete with a foreword, table of contents, and all cross-referenced articles. This is perfect for sharing findings with people outside your study room or publishing a finished investigation.
When you add entities from multiple sources — a Wikipedia extraction, a deep research report, manual entry — the same real-world person or place may appear more than once in your World. Notaepedia includes tools to find and merge these duplicates.
When you extract entities from a Wikipedia article, the system automatically runs fuzzy matching (Levenshtein distance + alias overlap) across all extracted entities. Entities with ≥95% name similarity are auto-merged. Those at 85–95% are flagged for your review. This happens before entities are saved to your World.
When you add an entity — whether from Wikipedia extraction, bookshelf promotion, or manual entry — the system checks it against all existing entitiesin your World using alias-aware fuzzy matching. If “Paul J. Fronczak” already exists, the system warns you before creating a duplicate. At ≥95% similarity, it skips creation entirely and maps to the existing entity.
For entities that already exist in your World, use the “Scan for Duplicates”button on your World's entities page. It runs pairwise fuzzy matching across all entities of the same type and shows you a review list. From there, you can merge two entities — the system automatically:
Each entity now tracks where it came from — a badge shows whether it was extracted from Wikipedia, promoted from a bookshelf document, created manually, or imported from a deep research report. When reviewing duplicate candidates, this provenance helps you decide which entity to keep as primary.
The Entity Graph gives you a visual map of your World. Every entity is a node; every relation is an edge. You can drag nodes to rearrange the layout, collapse sections, and zoom in on specific clusters.
The graph is especially useful for:
See the Entity Graph Tutorial for a complete guide.