How Workshop Works

Workshop turns intent into governed motion graphics. It is designed for situations where video output must be repeatable, inspectable, brand-controlled, and easy to revise.

The platform has five layers:

text
Intent -> .ws Source -> Compiler Gate -> RenderGraph -> Native Render -> Review/Publish

Layer 1: Intent

Intent can come from a human brief, a product workflow, a dataset, a source video, a template fill, or an AI agent. Intent is not yet a video. It is the business goal: what should be said, to whom, in what tone, with what constraints.

Layer 2: .ws Source

.ws is the authoring artifact. It is readable, diffable, editable, and suitable for both humans and agents.

text
Film "Product Launch"
  mood: precision
  aspect: 16:9

Scene "Problem"
  show "Teams need exact video, not lucky generations."
    size: title
    enter: rise

Scene "Solution"
  show "Workshop compiles intent into governed motion."
    size: hero
    emphasis: "governed motion"

The source expresses content and creative direction, not low-level pixels and keyframes.

Layer 3: Compiler Gate

The compiler is the authority boundary. It parses the source, resolves mood and theme policy, computes timing, solves layout, applies animation rules, and emits diagnostics when the artifact is invalid or risky.

This is what lets agents participate safely. A model can draft source, but the compiler decides whether the source is structurally valid.

Layer 4: RenderGraph

The RenderGraph is the resolved execution plan. It contains concrete text, timing, layout, color, animation, and media references. It is the protocol between authoring and rendering.

Because the RenderGraph is explicit, teams can inspect what will render before spending render time or publishing output.

Layer 5: Native Renderer

The native renderer executes the RenderGraph into MP4 output. Workshop's architectural goal is deterministic motion-graphics rendering without relying on browser capture as the production renderer.

The renderer exists to execute exact text, timing, typography, layout, media, and animation as specified by the graph.

Review And Publication

Rendering is not the same as publishing. A rendered artifact can still require brand, legal, compliance, customer, or operator approval. Workshop supports this by keeping the source, RenderGraph, MP4, metadata, and document history available for review.

The right question is not only "can the system generate a video?" It is "who decides the artifact is published, and what evidence do they review?"

Why This Matters

Prompt-to-pixels tools are powerful for exploration, but production teams need repeatability. Workshop provides:

  • Exact source-controlled text
  • Deterministic visual execution
  • Brand and mood constraints
  • Reusable templates
  • Inspectable RenderGraphs
  • Agent-compatible diagnostics
  • Human governance at the publication boundary

Next Steps