One canvas · Every phase of discovery

From raw evidence to shared direction.

Fabric brings research, diagrams, conversations, and decisions into one multiplayer workspace—then helps the team shape them without hiding the work.

Illustration · Evidence map

The system

Six capabilities, designed as one workflow.

Fabric is not another whiteboard with features attached. Each layer exists to move a team from scattered input to a decision it can explain.

Spatial freedom: Arrange notes, text, shapes, images, frames, and connectors without choosing a document structure first.
Structured when needed: Shared-board links pair a read-only visual canvas with an ordered semantic list of the saved board objects.
Live collaboration: Presence, comments, selections, and shared edits keep the team together without taking over local focus.
Local-first persistence: Fabric writes locally first, keeps visited work available offline, and merges authorized updates on reconnect.
Reviewable AI: The clustering skill proposes theme frames and moves as a visible preview. An editor applies or discards the patch.
Recoverable decisions: Named checkpoints, comments, and bounded history preserve context before a workshop or broad rewrite.

Wedge workflow

Keep the source material visible while the team moves from collection to synthesis, planning, and review.

01
Collect
Drop in screenshots, quotes, constraints, and rough notes.
02
Arrange
Cluster evidence spatially and connect ideas without flattening context.
03
Synthesize
Ask the clustering skill to propose theme frames and a clearer arrangement.
04
Decide
Review the patch, discuss exact objects, and checkpoint the outcome.

Honest scope

The current product concentrates on reliable editing, durable collaboration, accessible reading, and explicit AI approval.

Authenticated workspaces, member roles, durable boards, comments, share links, canvas editing, local recovery, and reviewable AI proposals.

Realtime persistence, reconnecting local queues, health probes, and isolated database identities run as explicit application services.

Multi-instance realtime fan-out, update compaction, independent security assurance, and contractual service levels are not asserted.

See how the pieces feel together.

Open your workspace, shape a board, and inspect every AI change before it becomes part of the canvas.