About this catalogue

Most online tools upload your file to a server, run it through a queue, and hand back a result. This catalogue doesn't — every instrument here runs entirely inside your browser tab.

How that's actually possible

Modern browsers expose the same kind of low-level capability that used to require installed software: cameras and microphones through getUserMedia, screen capture through getDisplayMedia, and — most importantly — near-native-speed code execution through WebAssembly. A video encoder, a PDF engine, an OCR model, even a speech-recognition model can be compiled to WebAssembly and run directly in a browser tab, at speeds that were desktop-only a few years ago.

A handful of tools go further and run real machine-learning models — background removal, image upscaling, speech transcription — using ONNX Runtime Web or similar in-browser inference engines, with GPU acceleration through WebGPU where the browser supports it, falling back to CPU-based WebAssembly otherwise. The model weights are downloaded once (and cached), but the actual inference — the part that touches your file — never leaves the tab.

What "instant" means here

Every tool page is usable the moment it loads — there's no heavyweight app shell to wait through. Anything genuinely large (a video codec, a machine-learning model) loads lazily, only when you actually start using that specific tool, with a visible size and progress readout rather than a silent spinner.

Why a catalogue, not an app

Each tool here is its own page, with its own URL, its own explanation, and its own interaction model — built like a catalogue of individual instruments rather than one monolithic app with a sidebar of features. That's partly practical (a tool is easiest to find by searching for exactly what it does) and partly a design choice: each instrument earns its own "face," suited to what it actually does, rather than being squeezed into one generic template.

Open source

The entire codebase — every tool's actual implementation, not just a description of it — is public at github.com/nabeel-ncz/tools. That's deliberate: a privacy claim is only worth as much as you can verify yourself. Read the source, run it locally, or fork it.

Something missing?

If there's a tool you'd expect to find here, request it on GitHub — new instruments get added based on what people actually ask for.