jarvisbox

jarvisbox /

AI tools

10 client-side AI utilities — OCR, image upscaling, background removal, speech, sentiment, summarization, face blur, keyword extraction, and QR decoding. Models load the first time you use them (5–30 seconds) and are cached locally. No API key, no upload, no signup.

100% client-side · no uploads · no telemetry

AI Tools

Why on-device AI?

Cloud AI services charge per request, log every prompt, and reserve the right to train on your data. Running the same models in your browser via WebAssembly, WebGPU, and the Web Speech API keeps your text, images, and audio strictly local — and after the first model download, the tools work offline.

From OCR and image upscaling to sentiment, summarization, and face blur, every tool here is built for the moment you need AI without the bill, the wait, or the privacy compromise.

常見問題

Do these AI tools upload my data?
No. Every AI tool on this page runs entirely in your browser using WebAssembly, WebGPU, or built-in browser APIs. Your text, images, and audio never leave your device.
Why is the first run slow?
Heavier AI tools download a model file the first time you use them (typically 1–140MB). After that the model is cached in your browser, so subsequent runs are fast. Lighter tools use built-in browser APIs and need no download.
Does my device need WebGPU?
No. Every tool works on CPU via WebAssembly. WebGPU, when available, makes some tools faster — but a CPU fallback is always provided.
回報這個工具的問題