ClouisleClouisle

15-Minute Quick Start

Sign in to Clouisle, choose a team, and complete your first Agent conversation

Clouisle is a multi-agent collaboration platform and workflow engine, with teams as the resource boundary. After completing this page, you will have opened a published Agent in your team workspace, sent a message, and found the matching conversation record.

Before you begin

You need an accessible Clouisle site and a valid account. Creating an Agent also requires a chat model authorized to your team; menus that regular members cannot see are determined by roles and permissions. Need your own site? Start it with Docker Compose: 4 GB RAM minimum, browser entry point http://localhost:3000.

Step 1: Sign in and choose a team

  1. Open the site root address and enter Login.
  2. Sign in with a username or email and password; the site may also offer SSO only.
  3. If you see prompts for email verification, TOTP, password expiration, or administrator approval, complete them on the page before continuing.
  4. Select the target team in the team switcher at the top of the workspace. Agents, Workflows, Knowledge Bases, and models are all isolated by the current team.
Selecting the current team
Selecting the current team

Step 2: Confirm available models

Go to Models. Only models authorized by administrators to the current team are shown here; you can filter by provider, model type, and status. You need at least one enabled Chat model to configure an Agent.

If the list is empty, contact the team owner or an administrator to authorize models for the team in the admin panel. This page is read-only: no user can add models or save provider credentials here.

Step 3: Open or create an Agent

  1. Go to Apps and switch to Agent.
  2. If you already have a published Agent, choose Chat from the ⋯ menu at the top right of its card, or open the orchestration page and click Chat in the toolbar.
  3. If you have the agent:create permission (the Admin and Member roles have it by default), select Create App, choose AI Agent under Select Type, enter a name, and enter the orchestration page.
  4. On the Orchestration page, write the Prompt; in Agent Settings > Model Configuration, select a model, then save.
  5. Publishing requires the agent:publish permission. Draft and published differ only in that publish state: iframe embedding works only while the Agent is published, while API calls just need a caller with access to the Agent and valid credentials.
Agent orchestration and preview
Agent orchestration and preview

Step 4: Send your first message

  1. Type a question in Debug & Preview on the right, or click Chat in the toolbar to open the Public Page in a new tab for the full conversation.
  2. Enter a message and send it. Streaming responses may include reasoning, knowledge sources, tool execution, and media results; control what is shown with Hide Reasoning and Hide Tool Calls in Agent Settings.
  3. To stop, select the Stop button in the input area. A manually stopped turn stays in its stopped state and produces no further output.
  4. Open Logs in the application sidebar to confirm conversations and messages, Token and duration statistics, and knowledge sources.
First Agent conversation result
First Agent conversation result

Acceptance

  • The current team is selected.
  • The Models page has at least one enabled Chat model.
  • The Agent status is Published, and it can return messages.
  • The conversation you just had can be found in the Agent Logs.

Next: learn to create and orchestrate an Agent or complete the RAG Agent tutorial.

How is this guide?

Using Clouisle

From first sign-in to building, publishing, and operating AI agent teams

Core Concepts

Understand Clouisle's resource model, how it runs, and governance boundaries

On this page

Step 1: Sign in and choose a teamStep 2: Confirm available modelsStep 3: Open or create an AgentStep 4: Send your first messageAcceptance