Overview
Eigent is a desktop app for handing work to AI agents. You can run one agent on a focused task or a team of agents that split a complex workflow and run in parallel. It's built on CAMEL-AI, supports any model, and can run entirely on your own machine. There are also cloud-connected, cloud and enterprise options.
Key capabilities
- Single agent or workforce: one meta agent, or coordinated multi-agent runs.
- Local deployment: local backend and local models, no cloud account required.
- Model agnostic: use any model, including vLLM, Ollama and LM Studio.
- MCP and skills: connect tools and add skills.
- Built-in browser and terminal toolkits.
- Automation: schedule tasks.
Tradeoffs
- Multi-agent runs use more tokens than a single agent.
- Local deployment needs a capable machine for local models.
- Enterprise features (SSO, access control) are a separate offering.
Ideal for
- People who want a local, open Cowork-style app
- Automating multi-step desk work
- Teams that need data kept on-premises
Not ideal for
- Simple one-question chats
- Users who don't want to manage models or a local server
Getting started
- Choose a setup: local deployment (recommended, no cloud account), cloud-connected quick start, or the hosted cloud version.
- For local use, follow the local deployment guide in the repository's
serverfolder and point Eigent at a local model server such as Ollama, vLLM or LM Studio. - Start with a single-agent task, then try a workforce run for a multi-step job.
- Add MCP tools and skills as you need them.
Technical notes
- Platforms: desktop app
- Based on: CAMEL-AI
- License: Apache-2.0
- Source: eigent-ai/eigent
- GitHub stars: 15,807 (as of 2026-11-12)