Skip to main content

​
What is Arden?

Arden is a local AI assistant that remembers useful context between conversations. It connects to your email, calendar, Slack, web, MCP tools, local files, and shell. Its home screen brings together conversations, requests, agent activity, and Areas: the parts of your life or work you want Arden to keep watching. Unlike stateless chat interfaces, Arden:
  • Focuses by surfacing work and requests that need you
  • Remembers conversations, commitments, and preferences across sessions, synthesized into readable topic pages
  • Acts autonomously via scheduled automations, standing area agents, background agents, and multi-agent workflows
  • Returns structured results: agent runs can produce validated objects, not just prose
  • Shows its work with streaming traces, approval cards, and tool/run lifecycle events
  • Renders rich output through sandboxed HTML widgets for charts, tables, forms, and visualizations
  • Integrates with Gmail, Google Calendar, Slack, web search, MCP tools, local files, and notifications
  • Keeps context lean by loading infrequent tool schemas only when needed
  • Runs locally – the server and clients run on your machine

​
Architecture

Arden is split into two main surfaces:
  • Server – Python/FastAPI backend for agents, tools, memory, automations, and the HTTP/SSE API
  • Desktop – Electron client for chat, areas, memory, automations, settings, approvals, and HTML widgets

Quickstart

Install and run Arden locally.

Concepts

How facts, wiki pages, areas, automations, and agents fit together.

Areas

Life domains, outcomes, requests, and standing agents.

Desktop Guide

Connect the app and use the main UI surfaces.

Memory

How persistent memory, topic pages, and recall work.

HTML Widgets

Render rich interactive cards from agent tools.

Automations

Schedule tasks that run autonomously.

Workflows

Subagents, multi-agent presets, and the five spawn surfaces.

API Reference

HTTP API for building on top of Arden.