NexusBond Vision Architecture Demo
Eyes
One clean interface. Multiple swappable AI vision models. The model is just a pluggable component behind the pipeline.
Model Output
Idle
Awaiting analysis
Upload an image, select a model, and run the same pipeline through different providers.
Implementation Stack
Vanilla UI, Node adapter layer, real vision APIs.
FrontendHTML, CSS, vanilla JavaScript
BackendNode.js HTTP server, single `/api/analyse` interface
RuntimeUbuntu VPS, systemd service, Nginx reverse proxy
ModelsGPT-4.1, Claude Opus 4, Gemini 2.5 Pro
SecurityServer-side API keys via `.env`, no browser exposure
PersistenceStateless demo, no database, no accounts