A workspace that remembers why
A shared visual system where spoken thinking becomes an inspectable product direction.
presence.render()voice → context → form
Voice will be the AI interface.
Visual context makes it usable.
Voice is the most direct way to give an AI a rich train of thought. But a long voice session becomes difficult to navigate, verify, and share when it disappears into a transcript.
Liveprompts explores a model where each instruction could remain linked to a canvas. Alone or with a team, concepts, decisions, dependencies, and revisions could retain shareable visual context.
format.select(best)
One voice stream.
Six illustrative views.
The prototype imagines how a visual language could preserve the meaning of speech and support the next decision.
Illustrative interface · Illustrative prototype data
01
Text
Keep the signal.
Long voice instructions could become concise briefs, decisions, and requirements—not an unmanageable transcript.
02
Diagram
Give complexity a shape.
Concepts, systems, people, and dependencies could take shape while the conversation is still unfolding.
03
Motion
Keep time visible.
Sequences and revisions could animate when the order of spoken instructions is part of the meaning.
04
Video
Share the direction.
A voice session could become a visual narrative when tone, rhythm, and atmosphere carry the idea.
05
Micro app
Turn intent into a tool.
Spoken requirements could become illustrative trackers, controls, and prototypes.
06
3D scene
Make architecture tangible.
Products, spaces, and systems could become illustrative 3D models that clarify scale and relationships.
conversation.without_the_noise()
An envisioned interface for sustained voice-first work with AI.
Speak at the speed of thought.
The concept explores an interaction that could respect pauses, restarts, and changes of direction without imposing prompt syntax.
Imagine every instruction in context.
The prototype illustrates how each part of the canvas could retain a visible link to the instruction that created, related, or revised it.
A conversation the whole team can see.
The canvas could become shared memory: inspectable, editable, shareable, and preserved through every version.
presence.merge(real_time)
A canvas for each person.
One shared train of thought.
The envisioned interface would give each participant a personal space while keeping a shared canvas visible.
Prototype interaction · Illustrative scenario · fictional persona · Illustrative prototype data
illustrative collaborationROOM / 03 · ILLUSTRATIVE
AMJLAI
personal canvasAMthinking
IDEA / 04Make trust visible
working hypothesis
voice 01send to shared →
shared canvas3present
personal canvasJLthinking
IDEA / 07Connect the system dependency
working hypothesis
voice 02← pulled from shared
Think in your own space.
The scenario imagines a voice-led thread explored privately before it is shared with the group.
Let the idea move.
Send a card, decision, or architecture into the shared canvas—with its source and context intact.
Build shared memory.
The scenario illustrates how a group could follow the origin of each idea and its place in the wider system.
canvas.open(what_matters)
Speak at full depth.
Keep the idea shared.
Turn spoken product thinking into a shared specification.
Vision, requirements, system maps, prototypes, decision history.
Explore a complex idea without losing the thread of the conversation.
Visual anchors, open questions, decisions, and evolving context.
Explain a system out loud without losing its structure.
Components, dependencies, flows, scenarios, and interactive models.
experimental prototype · work in progress
Explore the concept.
Try the interaction.
Liveprompts is a conceptual prototype. No commercial service or product access is currently offered.
Try the prototype