Nodes
Every AI output becomes a persistent card on an infinite canvas. Text, code, data, visuals, documents: 15 node types, each one reusable.
Spatial canvas for AI work
The persistence layer for AI work.
Every answer stays, connects, and becomes context for the next question. Outputs land as connected, reusable nodes. Thinking compounds instead of scrolling away.
Works with Claude, GPT and Gemini.
Claude · text
Interview themes
GPT · data
Survey results
Gemini · doc
Lit review notes
Synthesis · claude
contradictsinformed bysynthesisesDraft framing
sonnet3 in context
What it does
Chats scroll away. Canvases build up.
Edges
Draw labelled relationships between ideas: informed by, contradicts, synthesises. The map of your thinking stays visible.
Context
Select any nodes and the AI sees them. Your canvas is the conversation's memory, not a transcript you have to re-explain.
How it works
A loop, not a scroll.
01
Chat
Ask Claude, GPT or Gemini anything, right on the canvas.
02
Materialise
The answer lands as a node. It stays where you put it.
03
Connect
Draw labelled edges. Group related nodes into sections.
04
Re-ask with context
Select nodes and ask again. The AI reads what you selected.
Work together
Your canvas is shareable now.
Share links
Publish a read-only snapshot of any canvas and send the URL to anyone. No account needed to view it.
Live presence
See who else is on the canvas, cursor and all. Two people can think on the same surface at the same time.
Invite editors
Add people to a canvas by email. Viewers read, editors work alongside you, and you stay in charge of the list.
Who it is for
Built for research synthesis, literature reviews, product thinking and anyone whose AI chats are worth keeping.
ClaudeGPTGemini