Hushscript

Hushscript turns audio and video recordings into clean, speaker-labeled transcripts entirely in the browser — on desktop, tablet, or phone. There's nothing to install and no local AI models to configure.

It's made for people who work with sensitive recordings — interviews, meetings, medical and legal calls — and don't want a subscription or a meeting bot sitting in every call. For video, only the audio is extracted on your device and sent for transcription; the video file never leaves your computer, and the audio is deleted as soon as the transcript is ready.

Hushscript is our own product — designed, built, and operated by the same team behind LotusWebAgency. It's the clearest example of how we think about privacy, performance, and product.

The backend is written in Rust and compiled to WebAssembly. Rust gives us memory safety without a garbage collector, so the code that handles your recordings stays fast and predictable and sidesteps a whole class of crashes and security bugs — and it runs at the edge, close to the user, with near-instant startup.

The app is TypeScript from top to bottom and shares typed contracts with that Rust backend. Change an endpoint on the server and the client stops type-checking before anything ships — so refactors are safe, integration bugs surface at compile time instead of in production, and a small team can move quickly without breaking things.

The public site is completely static — pre-rendered HTML with no server round-trip and no framework to boot before content appears. That means near-instant loads, strong Core Web Vitals, and pages search engines can read in full on the first crawl. It's the same static-first approach behind the site you're reading now.