Rejoining... Retrying in s Connection lost Session paused Resume failed
Ariadne User Documentation
Documentation for how to use the Ariadne platform

getting-started

Last updated 13 September 2026

Getting Started with Ariadne

Ariadne is a continuous cognition platform that lets you work with AI agents across conversations, memory, tools, and integrations. This guide explains the core concepts and helps you orient yourself in the interface.

Accessing Ariadne

Ariadne can be used in several ways:

Client Best for
Web UI Everyday use from any browser
Desktop App Windows users who prefer a native experience
Console App Developers, headless environments, and IDE integration

The Web Interface

After logging in, the web UI is organised into four top-level menus.

Home

Your main working area. Contains all conversation and collaboration surfaces:

  • Chat – General-purpose AI assistant for questions, analysis, writing, and reasoning
  • Code – Coding-focused conversation mode
  • Query – Data analysis with structured data querying
  • Briefing – Automated personalized briefings
  • Notes – Markdown notes with AI-powered search
  • Whiteboard – Collaborative visual workspace
  • Visualise – 3D object scene management
  • Work – Kanban project and task tracking
  • Projects – Project workspaces: charter, conversations, work, documents, code, releases and support in one place

Tools

Standalone productivity tools you operate directly:

  • View Mailbox – Browse connected email accounts
  • View Calendar – See upcoming calendar events
  • Transcribe – Convert audio to text
  • Read Content – Have content read aloud (text-to-speech)
  • Deep Research – Multi-round, multi-source research producing a cited report
  • Summarisation – Summarise URLs, videos, audio, and documents
  • Web Spider – Crawl and index website content
  • Website Automation Flows – Record and replay browser automations

Settings

Configure how the platform behaves. Typically visited by administrators or power users:

  • Agents – Manage AI agent configuration, prompts, and tool collections
  • Personas – Create public-facing AI personas with embeddable chat widgets
  • Guard Rails – Define safety and content governance rules
  • Knowledge – Build knowledge bases agents can draw upon
  • E-Mail Config – Connect email accounts
  • Calendar Config – Connect calendar accounts
  • Workflows – Build workflow automations
  • Goals – Define long-term objectives for agents
  • Schedules – Create time-based automation triggers
  • Webhooks – Set up external event triggers
  • Tools – Manage tool collections per agent
  • MCP Servers – Connect external tools via Model Context Protocol

User Menu (top-right)

Identity and account management:

  • My Profile – Edit your display name and preferences
  • Authorise Access – Generate API tokens for programmatic access
  • My Organisation – Manage your organisation and invite members
  • My Subscription – View and manage your subscription
  • My Voice Accessories – Configure voice input/output devices
  • My Physical Accessories – Set up physical device integrations
  • My Email Addresses – Manage connected email identities
  • My Calendar – Manage calendar account connections

Core Concepts

Conversations

Every interaction with an agent takes place inside a conversation. Conversations persist indefinitely — you can return to any previous conversation and continue where you left off. Each conversation has its own history, context, and optionally its own set of tools.

Conversation modes shape the agent's behaviour and default tool set:

  • Chat – balanced, general assistant
  • Code – prioritises software development tools
  • Query – focuses on data analysis and structured outputs

Agents

An agent is a configured AI persona with a system prompt, tool collection, reasoning settings, and optionally connected MCP servers and knowledge bases. You can create multiple agents for different roles (e.g. a research agent, a coding agent, a customer support agent).

Memory

Ariadne stores all conversation history and can retrieve relevant past context using semantic search powered by vector embeddings. This means an agent can "remember" things from earlier conversations when they become relevant.

Tools

Tools extend what an agent can do beyond pure conversation. Examples include:

  • Searching the web
  • Reading and writing files (in Console/headless mode)
  • Querying databases
  • Running automation flows
  • Sending and reading email

Tools can be grouped into collections and assigned per agent. See Agents for details.

MCP (Model Context Protocol)

Ariadne connects to external MCP servers to gain additional tools, resources, and prompt templates from third-party services. Ariadne can also act as an MCP server itself, letting external tools like Claude Desktop access your notes, knowledge, and conversations. See Agents → MCP Servers for configuration details.

First Steps

  1. Start a chat – Go to Home → Chat, type a message, and press Enter.
  2. Set up an agent – Go to Settings → Agents to customise the system prompt and tool collection.
  3. Add knowledge – Go to Settings → Knowledge to create agent skills or dynamic facts.
  4. Configure email – Go to Settings → E-Mail Config to connect your mailbox.
  5. Create a persona – Go to Settings → Personas if you want a public-facing widget on your website.
  • The Home menu is where you do most of your work
  • The Tools menu contains things you operate directly (summarise a URL, view your inbox)
  • The Settings menu is where you configure how the system behaves
  • Conversations are listed in the sidebar on the left; click any to open it