CLASSEVE
RouteHome
CLASSEVE

Research that becomes working software.

We build resilient systems, run our research in the open, and back every claim with proof. No fluff — just software that works.

The flagship · for Windows

REX runs your computer for you.

Ask by voice or by typing. REX opens your apps, works through the steps, and gets it done — on your machine. It’s the main thing we’re building.

Expected launch: August 4, 2026 — the waitlist hears first.

Releases

Working Software.

Current shipping and early-access builds. Open a product for its actual platform status, requirements, and known limits.

Beta

Lven Instant

Instant, on-device voice-to-text — Windows, Linux, and Android today; macOS and iPhone soon.

Speak, and the text appears the moment you stop talking — in whatever app has focus. Transcription runs entirely on your device: no audio upload or inference round trip. Model setup, account validation, and token refresh still use the network. Available now on Windows, Linux, and Android, with macOS and iPhone on the way.

WindowsLinuxAndroidmacOSiPhone
Operational viewLven Instant
TranscriptionInstant
Your audioStays local
Platforms live3
Beta

Lven Cloud

Server-side voice-to-text tuned for the highest accuracy, on every platform.

The same instant dictation, run on a dedicated server engine for the highest accuracy we can deliver on any hardware. Hold the hotkey, speak, and polished text lands at your cursor in well under a second — on Windows, Linux, and Android today, with macOS and iPhone on the way.

WindowsLinuxAndroidmacOSiPhone
Operational viewLven Cloud
Transcription<700 ms
AccuracyHighest
Platforms live3
Beta

earslate

Free live speech translation in your ear, on Android.

Walk into a room where nobody speaks your language and hear them as if they did. earslate streams nearby speech directly to Gemini or OpenAI and plays the translation back in your earbuds in real time.

Android
Operational viewearslate
LatencyLive
PriceFree
Providers2
Frontier

Active Research.

Unresolved problems, in-progress builds, and open edge cases. We document the work in the open, including what's still unsolved.

Exploring

Lven Web

Browser-based voice transcription without installation.

A web interface for Lven that works directly in the browser. No download, no install. Open the page, speak, copy.

Read frontier detail
Proof

Measured Reality.

Documented performance, hard limits, and architectural constraints. Trust requires actual data, not marketing claims.

Cross-platform, on-device everywhere it ships

One Lven Instant subscription covers every platform as it ships — Windows, Linux, and Android today, macOS and iPhone on the way. Every shipped build runs the same on-device transcription engine.

Each platform uses its native audio stack and text injection method. No emulation layers, no shared web shells.

May 20, 2026Related route

Transcription is instant

Speech becomes text on your device with no upload step and no network round trip. The result appears the moment you stop speaking.

The quantised speech model is downloaded during setup and then runs on your CPU. No remote inference means the transcription path has no upload queue, server rate limit, or network round trip.

May 20, 2026Related route

Overlay stays stable across sessions

The floating bubble survives window manager changes, display sleep, and app switching.

Qt6 window flags ensure always-on-top behavior. Position persists across recording sessions.

Apr 5, 2026Related route

Local-first recording

Audio is written to disk before transcription. Network failure does not lose the recording.

WAV files persist in the cache directory until transcription succeeds or the user clears them.

Apr 5, 2026Related route

Text lands at cursor position

Transcribed text is injected into the active input field at the exact cursor location.

On Windows, Lven tries UI Automation and falls back to a controlled clipboard paste. Linux currently uses xdotool on X11 or wtype on Wayland. macOS and iPhone remain unshipped, and Android uses its current platform service.

Apr 5, 2026Related route
02 - ProofOpen all proof05 entries archived
Standard

The Baseline.

The engineering gate used to distinguish production-ready software from early access and research.

What reaches release

Production-ready status is reserved for software that operates correctly under sustained real-world load.

The primary action is obvious within the first minute of use.Failure recovery is engineered before the first public session.
What stays in research

Frontier is where active research lives. We document the work in the open and label exactly where it stands.

A direction is established, but technical edge cases remain unresolved.State labels match the current build, not the target build.
What goes public

Public means the source or download is available. It does not automatically mean production-ready or suitable as a security boundary.

They solve one specific, well-defined problem.Their product page states requirements, maturity, and known limits.
What gets rejected

Some builds run correctly but still fail the baseline. Function alone is not sufficient.

Interfaces that hide state behind decoration.Systems that need explanatory copy to clarify their primary function.
04 - StandardRead the standard04 rules in force
Select

Selected Work.

Everything we make in one place — the products we sell and the public builds we give away, each held to the same engineering baseline.

05 - SelectView the catalog06 in the catalog