Quick Start

Prototype Documentation: UPAS is currently in early development. This guide describes the intended user experience for the planned production release.

Prerequisites

Before you begin, ensure you have:

A modern web browser (Chrome, Edge, or Safari recommended for WebGPU support)
Sufficient device storage (500MB–2GB depending on model and pack sizes)
Initial network connectivity to download assets (not required after caching)

Using the PWA

UPAS is distributed as a Progressive Web App (PWA). To get started:

Navigate to UPAS

Open the UPAS application URL in your browser. On first load, the service worker will begin caching static assets.

Wait for Caching

The application will display progress as it downloads and caches:

Application shell and static assets
AI model artefacts (GGUF weights)
Configured procedure packs

Install as App (Optional)

Most browsers will offer to "Add to Home Screen" or "Install App". This provides a native app experience and faster launch times.

Query Guidance

Once caching completes, you can query guidance even without connectivity. Type your question and receive stepwise guidance with source attribution.

Runtime Detection

UPAS automatically detects your device's capabilities:

// Automatic runtime selection
if (navigator.gpu) {
  // WebGPU available: fast, GPU-accelerated inference
  runtime = "webgpu";
} else {
  // Fallback to WASM: works everywhere, slower
  runtime = "wasm";
}

A badge in the UI indicates which runtime is active:

WebGPU: Optimal performance, GPU-accelerated
WASM: Broad compatibility, CPU-based
Manual: AI unavailable, browse procedures directly

Offline Verification

To verify offline capability:

Load UPAS and wait for caching to complete
Enable airplane mode or disconnect from the network
Refresh the page — it should load from cache
Submit a query — guidance should be returned locally

The service worker uses a cache-first strategy for model artefacts and application assets. Network requests are only made when content is not in the cache.

Next Steps

Installation — Deployment options
Configuration — Model and pack settings
Guides — Detailed workflows

Quick Start

On this page