Rich media
Inspect images, audio, video, and text beside the knowledge they describe.
+ Explore Cronologies
Explore
your data.
See it from every angle. Map meaning, search across media, build connections, and move between views without losing useful context.
MEANING → DISTANCENODES + LINKS01 Semantic maps
Meaning becomes
distance.
Embeddings are computed right in your browser and projected into 1D, 2D, or 3D layouts. Text, images, and video scenes share one space, so meaning lines up across media—related records drift together and outliers stand apart. What was a table becomes terrain.
- SEMMap by concept
Use meaning as spatial distance, not only as labels or columns.
- MAPProject many dimensions
Use UMAP and layout controls to turn complex similarity into navigable space.
- MMMove across media
Compare meaning carried by text, imagery, video scenes, and other embedded material.
0.94 Nearby concepts
0.18 Outlier cluster
02 Viewpoints & ontology
The same universe,
seen differently.
Lay the same universe out however the question demands. Place nodes by a coordinate—a property value or a UMAP-reduced embedding—or let a force-directed layout arrange them by their connections. Map any field to position, size, or color.
- POSPlace by value
Position and color nodes by a property value or a UMAP-reduced embedding coordinate, in 1D, 2D, or 3D.
- FDLLay out by force
Switch to a force-directed graph so connected nodes pull together and clusters emerge from structure, not coordinates.
- EDGDraw the edges
Add typed connections by hand between people, concepts, places, and media—and watch the layout rearrange.
- ONTBuild an ontology
Create dimensions, name values, and define parent-child conceptual hierarchies.
03 Many views, one universe
Orbit it. Walk it.
Play it through time.
Move between globes, grids, timelines, and 3D graphs without leaving your data. Drop to first person and walk the surface, or press play and watch your universe evolve over time.
- VUESwitch forms
Move between globe, timeline, graph, similarity, first-person, and inverted views.
- FPVEnter the field
Use first-person and Tree of Knowledge views to experience structure from within.
- TIMPlay time forward
Use temporal controls to watch records appear, cluster, and change.
04 Cross-modal search
Search by word,
or by example.
Because text and media share one space, a phrase can surface images and an image can surface text. Query with words or an example, combine several items into a single query, and narrow the result with faceted filters.
- TXTSearch in words
Use plain language; matches come back by meaning, not just exact keywords.
- IMGSearch by example
Drop in an image or a record and find the closest images and text alike.
- CTRCombine into a centroid
Select several items and search from their shared center to find the space between them.
- FILFilter the region
Layer facets, dimensions, ranges, and categories on top of any search.
this image + “mythic machinery”
0.91 Image match
0.86 Text match
05 Media understanding
Read what's
inside the media.
Before anything is mapped, your content is broken into meaningful pieces: entities pulled from text, long documents split into passages, and video cut into scenes. Each piece is embedded on its own, so you can search into the middle of a document or the exact moment in a clip.
- NERRecognize entities
Pull people, places, organizations, and concepts out of documents automatically.
- SEGSegment long text
Split documents into passages so each idea is embedded and found on its own.
- SCNCut video into scenes
Detect scene changes and turn each shot into its own searchable moment.
- VIDSearch inside video
Jump straight to the scene that matches a phrase or an example image.
06 Create a universe
From mixed data
to explorable worlds.
Import mixed source material, split it into meaningful pieces, embed those pieces, extract the people, places, objects, and concepts they refer to, then move through the resulting graph as links or semantic neighborhoods.
07 Private by design
Your data
stays yours.
Import CSV, XLSX, JSON, PDF, and media files. Most parsing, entity extraction, and embedding happen locally in your browser. Large CSVs can use account-based cloud processing when server processing is enabled.
- IMPBring many sources
Start from spreadsheets, structured files, documents, images, audio, or video.
- LOCLocal by default
Build and refine most universes in browser storage; cloud-processed datasets remain tied to your account.
- EXPCarry it forward
Export bundles that preserve media, edits, routes, and structure.
LOCAL ONLY READY
+ More ways to work
Bring the whole
universe with you.
Explore deeply, add what you learn, and carry the result into the next investigation.
Connected navigation
Travel from a selected node to the people, ideas, places, and media around it.
Collections
Gather significant nodes into working sets without losing their context.
Browser editing
Refine metadata, relationships, dimensions, and ontology as you learn.
Compare selections
Hold several nodes side by side to weigh their fields, media, and neighbors at once.
Saved routes
Record a path through knowledge and replay the sequence for someone else.
Begin anywhere
The next connection
is already there.
You only need a different way to see it.