Skip to content

HomeDownload

Run it on your own machine.

Cognia is a desktop application. When a release is published, the installers below come from a signed build of the public source.

Build from sourceView source

No installer has been published yet. The build is public and reproducible — clone the repository and run the desktop build, or watch releases to be notified when the first one ships. Watch releases

Detecting your platform

Why the workbench runs on your machine.

The desktop shell is not packaging around a web page. It is what lets one task reach local projects, native services, portable data, and another authenticated device.

  • Native desktop shell

    A Tauri host connects the static interface to Rust services for files, processes, automation, and platform integration through explicit commands.

    Tauri · Rust services · command bridge

    Learn more

  • Project tools in place

    Use the integrated terminal, editor, source control, and project instructions against the files already on your machine.

    Terminal · editor · source control

    Learn more

  • Portable local data

    Export and restore conversations, settings, memory, and artifacts through a documented backup format instead of depending on an account silo.

    Backup · restore · export

    Learn more

  • Connected companion

    Reach the running desktop workspace from an authenticated companion device through the same service plane and explicit connection controls.

    Companion · authentication · live session

    Learn more

Build from source

The desktop build is reproducible from the repository. This is the same path the release workflow takes.

  1. 01 · Clone the repository

    git clone https://github.com/MaxQian888/cognia-next

  2. 02 · Install dependencies

    pnpm install

  3. 03 · Build the desktop application

    pnpm tauri build

What you need

  • Node.js and pnpm, to install dependencies and build the frontend.
  • A Rust toolchain, because the desktop shell is compiled rather than bundled.
  • A model to talk to: a local runtime, a provider key, or a subscription you already have.