Layer
Stream Protocol
A typed event contract for messages, widget creation, incremental HTML frames, component payloads, and host callbacks.
Platform
A complete stack for streamed AI interfaces.
StreamCanvas is not just a visual demo. It is a delivery model for interactive AI experiences with typed streams, safe rendering, component registration, and self-hosted operational control.
Layer
Rendering Layer
A sandboxed iframe renderer for HTML widgets plus schema-validated React components for app-native surfaces.
Layer
Action Bridge
Client tools, approvals, and structured payloads let widgets trigger trusted host actions without arbitrary DOM access.
Layer
Operational Delivery
Reference server helpers, standalone runtime support, and reverse-proxy deployment patterns for self-hosted teams.
Why teams use it
Keep the interaction model stable while the implementation matures.
Teams usually start with a high-visibility demo, then discover they need approvals, persistence, local actions, and deployment discipline. The platform exists so those requirements can be added without abandoning the user experience that made the product valuable in the first place.
01Sandboxed widget rendering by default
02Schema-validated React component mode
03Host-only actions through explicit client tools
04Self-hosted deployment on localhost-bound services
Deployment modes
Local development - pnpm dev Production deployment - Next.js standalone runtime - systemd service - nginx reverse proxy - localhost-only app port
Next step
Review the operational pages that explain deployment, security, and how teams apply the platform to customer-facing AI experiences.