Skip to content
Browse documentation

Getting started

Install, sign in, and run a first request

Set up a licensed desktop workspace and safely verify a small task.

Vopit App coordinates an editable team of AI agents in your local project. Install an available release from Downloads, sign in with an active Vopit license, then configure a model connection. Your Vopit license and model-provider access are separate.

Install Vopit App and open a project

  1. Open Downloads and choose how to install: the Microsoft Store, the Windows installer, or one command in PowerShell (irm https://vopit.fr/install.ps1 | iex), which also adds the vopit command to your terminal. On Linux with apt (Debian and the distributions based on it), download the .deb package or run curl -fsSL https://vopit.fr/install.sh | sh. See Use Vopit from the terminal.
  2. Install the downloaded application and open Vopit. Choose Sign in to Vopit, complete the browser flow on the same computer, and return to the app. Before you sign in, the Settings button on this screen opens Account, Updates and Network settings, for example to add a corporate network certificate.
  3. Choose Open a project and select the local folder. A new project asks how to start its team; see the next section.
  4. Open Settings → Connections, choose a connection and model, and save before submitting work. A fresh project's default connection is Ollama with no model selected, so it cannot run a request yet. Until the project has a team, Settings shows only Account, Connections, Appearance and Network. A model catalogue loading successfully does not prove that the provider will accept an inference request.

Choose how to start the team

OptionWhat happens
Create your team with AI → Build my studioDescribe your activity, goals and the apps you use. Optionally enable Read this project to fit the team to it: README files, agent instruction files, manifests, docs and folder/file-type counts are sent to the selected model, with secrets masked; .env and key files are never read. Review the preview of teammates and suggested Crank automations, then choose Create this studio. Suggested Cranks are created disabled and nothing runs automatically. This needs a configured connection and model.
Create a base team → Create teamCreates the ten-agent native team. Orchestrator, Planner, Researcher, Developer and Reviewer start enabled; UI/UX Designer, Documentator, DevOps, Ticketer and Git Manager start disabled, ready to enable.
Import a team → Import teamThen choose Choose a team file (an exported team), Choose a source project (copy the team of another project folder), or Recover previous workspace, shown only when a previous workspace is available. Imports bring agents, Cranks, furniture and rooms; credentials are not included.

Run a bounded first request

  1. Choose a disposable project folder. Do not use a production repository for your first run.
  2. Confirm that the project team has an enabled coordinator and an enabled implementation agent. Keep the header speed mode on Big brain, the default, so planning and review run as configured.
  3. Ask for a small, observable result such as: “Add a short README section describing the local test command. Do not change application code. Report the file changed and the command you ran.”
  4. Review every permission request, then inspect the result and the changed file yourself.

All documentation