Home / Products / wAIve Desktop

The whole agent. On your computer. With any model.

wAIve Desktop is the wAIve agent as a desktop app. Your files stay in a folder on your disk, and you choose which model does the thinking.

Connect anything in one box

Type an IP, host:port, URL or provider name. It finds the server, lists the models, reads the context window and runs a test. Works with local servers like Ollama, LM Studio, vLLM and llama.cpp, the major cloud providers, and the mTec Engine.

A folder is the workspace

Plain files on your disk. Every write is versioned and deletes can be restored. Set each chat to Auto, Ask or Locked.

Works with small local models

Pinned notes, smart forgetting, and it never over-fills the context window. Built so a 30B model on your own GPU can finish a long task.

Tasks that finish

Auto-continue, an honest Done / Remaining pause, resume, self-checks and loop guards.

Sandboxed code

Runs code in Docker, bubblewrap or the macOS sandbox. Plain truth: on Windows without Docker, code is not isolated.

Always on

Lives in the tray. Scheduled tasks and notifications keep working while you do something else.

Set up in four steps

01

Connect your AI

A local server, a cloud provider, or the mTec Engine.

02

Pick a folder

That folder becomes the agent's workspace.

03

Choose a sandbox

Docker, bubblewrap or macOS sandbox.

04

Add services

Optional: search, voice, images, schedules.

Reads the web for less

CTX built in

Web pages come in as CTX, with ~85% fewer tokens than raw HTML. That matters most on small local models with small context windows.

Download

Windows installer, signed

wAIve Desktop ships as a signed Windows installer. macOS and Linux build from source.

Get wAIve Desktop

Every mTec hardware box ships with wAIve Desktop installed.

Pointed at the box, ready for your team on day one. Your models, your building, your files.

Start with a free exploration call.

Tell us what you're trying to do and what's in the way. If AI fits, we'll show you where. If it doesn't, we'll say that too.