orbix tech
All work

Open source

Artemis

An open-source AI agent for the desktop. It works in your files, terminal and its own browser, runs on any model, and asks before it acts.

Result
70 built-in tools, as many agents in parallel as you want, free under AGPL-3.0
Stack
Electron, React, Vite, SQLite, MCP
Year
2026
Licence
AGPL-3.0, free
Platforms
macOS (Apple silicon) build; Windows and Linux from source
Made by
Orbix Lab
Artemis screenshot

Not a chat window glued to one API. An agent that works where you work.

Does the work

Reads, writes and edits project files, runs commands in your real shell, searches the web and drives its own browser with saved logins.

Any model

OpenRouter, OpenAI, the Anthropic API, Cloudflare, Ollama, LM Studio or any OpenAI-compatible server, with daily and monthly spend limits.

You press approve

File changes, terminal commands and computer control show an approval card first. Deleting a file always asks, even with auto-allow on.

Agents and Brains

Run as many agents in parallel as you want, each with its own role, and named Brains on different models can answer alone or vote as a council.

Memory in plain files

Chats live in a local SQLite database and memory in Markdown files on your disk. No analytics, no crash reporting, no account.

Runs while you are away

Scheduled tasks, a heartbeat checklist and webhooks from GitHub, Stripe or n8n start runs, and results can go to your phone over Telegram.

Voice and app lock

Sentinel listens for a wake word with speech recognition on your machine. The app locks behind a password, Touch ID or your own voice.

MCP both ways

Use MCP servers as connectors, or let Claude Code and other clients drive Artemis through its local MCP server, gateway API and CLI.

Artemis running a task in a chat
A task in progress
Quick window opened from the tray
Quick window from any app
Sub-agents working in parallel
Sub-agents working in parallel
Notes with wiki links and a graph view
Notes, memory and a graph

Artemis is independent and not affiliated with Anthropic or any other model provider. Setup guides for every feature are in the wiki.

Tell us the task your team is tired of doing by hand.

A 30-minute call. We map the process, say whether it is worth automating, and send a fixed quote if it is.