Architecture

Functional flow

flowchart TD
    U[Field engineer] --> C[Site configuration UI]
    C --> P[Port availability model]
    C --> I[Accessory inventory model]
    P --> R[Routing engine]
    I --> R
    R --> S[Splitter selection]
    S --> A[AI PC connection]
    A --> M[Monitor allocation]
    M --> G[Optional Gateway allocation]
    G --> V{All required paths valid?}
    V -- No --> B[Rollback candidate state]
    B --> R
    V -- Yes --> D[Mermaid topology]
    V -- Yes --> O[BOM and checklist]

Core state

The routing engine tracks:

• remaining accessory inventory
• consumed-item BOM
• Mermaid graph source
• remaining AI-PC outputs
• defined graph nodes

A candidate route is evaluated against the current state. If a downstream connection becomes impossible, the previous state is restored and another candidate is tried.

Routing layers

Raw-video distribution

Select a splitter that supports the required number of raw-video branches.

Endoscopy output normalization

Use direct cables or converter chains when the endoscopy output differs from the selected splitter format.

AI processing PC input

Connect the raw feed to the selected AI-PC input with available materials.

Monitor routing

In one-monitor mode, preserve both raw bypass and AI output through either an external switch or separate monitor inputs.

In two-monitor mode, allocate one monitor to AI video and one to raw video.

Gateway routing

When enabled, Gateway input can originate from either AI output or raw video. Gateway output is then matched to a compatible monitor input.

Production evolution

1. Move equipment and accessory definitions into versioned JSON.
2. Represent connectors and converters as a directed compatibility graph.
3. Use weighted path search to minimize adapters and signal degradation.
4. Add unit tests for verified equipment combinations.
5. Add offline-first PWA support.
6. Add signed installation worksheet export.