Experience renderer¶
The experience renderer is one modality adapter in the Unison Surface. It composes a native visual and keyboard experience from the same governed semantic outcome available to conversational, Braille, tactile, sign-language, BCI, and future modality adapters.
The browser renderer is also a practical development and evaluation surface. Its interface remains a visual experience. Blind and visually impaired people receive native conversational or Braille experiences through their applicable adapters.
Responsibility boundary¶
The renderer can:
- express meaning, provenance, uncertainty, choices, confirmation, cancellation, and recovery from a semantic outcome;
- adapt visual structure, density, language, contrast, motion, and keyboard interaction to the person and situation;
- collect input and return a proposed intent, selection, or confirmation; and
- maintain presentation state for the current visual experience.
Identity, consent, policy, incident state, context access, tool authority, and physical actuation belong to their owning services. A renderer presents an authorized decision and relays the person's response. It cannot grant itself access, reinterpret a confirmation, or execute an action.
Common integration point¶
Every modality adapter connects through the versioned semantic experience contract. The contract carries the meaning and controls that must remain equivalent across native expressions:
- interpreted intent and current state;
- facts, sources, relationships, and uncertainty;
- available choices and actions;
- disclosure and confirmation requirements;
- cancellation, interruption, and recovery; and
- content-minimized receipt information.
A visual composition can use spatial comparison and direct manipulation. Conversation can use dialogue, pacing, and auditory structure. Braille can use tactile regions and direct navigation. New modality contributors implement the same contract through the strengths of their modality.
Development surface¶
With the development stack running, the visual renderer is available at
http://localhost:8092. It exchanges semantic outcomes and proposed person
responses with the control plane through versioned interfaces.
Renderer acceptance includes semantic contract tests, keyboard operation, reflow, zoom, high contrast, reduced motion, error recovery, cancellation, and confirmation fidelity. Cross-modality golden journeys verify equivalent meaning and control. Representative devices and participatory evaluation provide the evidence required for a supported native experience.