How to Deploy a 24/7 AI Assistant with Moltbot and MiniMax M2.1

In early 2024 the open‑source project Clawdbot, now renamed Moltbot, attracted massive attention on GitHub, quickly reaching over 66 K stars. Moltbot is positioned as a 24/7 AI employee that lives inside chat applications (Telegram, WhatsApp, iMessage, etc.) and can act on your behalf by accessing GitHub, Gmail, calendars, and even performing actions on your computer.

What Is Moltbot?

Moltbot is an AI agent that initiates interactions ("AI finds people") rather than waiting for a user request. It can read and write files, send emails, modify code, and retain a persistent memory of recent conversations, storing data locally for up to two weeks.

Installation

The installation is a single‑line command that works on a remote server or a local machine: curl -fsSL https://molt.bot/install.sh | bash After the script runs, you will be prompted to accept the risk (choose "Yes") and then select the "Quick Start" configuration mode.

Choosing a Large‑Language Model

Moltbot requires a backend LLM. Besides the traditional Claude models, the guide recommends MiniMax M2.1 as the most popular and well‑adapted choice. Obtain an API key from the MiniMax website and paste it when prompted.

Telegram Integration

To use Moltbot with Telegram:

Open Telegram and search for @botfather.

Send /newbot, give the bot a name, and copy the generated token.

On the server, run clawdbot pairing approve telegram <pairing‑code> (the pairing code is shown in the Telegram chat after sending /start to the bot).

Verify the service is running, e.g. ss -lntp | grep 18789. If no output appears, create a systemd service.

Feishu (Lark) Integration

For users in China, Feishu offers an easier entry point. Clone the community connector repository:

git clone https://github.com/m1heng/Clawdbot-feishu

Install the plugin inside Moltbot: clawdbot plugins install @m1heng-clawd/feishu Then create an application on the Feishu Open Platform (open.feishu.cn) to obtain an App ID and App Secret, and send those credentials to the Moltbot chat interface.

Running Moltbot

Start the gateway in verbose mode to see the listening port (default 18789): clawdbot gateway --verbose Once the service is up, you can interact with the AI assistant through the chosen chat client. The MiniMax M2.1 model provides fast, stable responses and handles long‑chain tasks such as price comparison for car dealers, automated email handling, insurance claim processing, flight booking, and more.

Conclusion

After configuration, Moltbot runs continuously on a machine that can stay powered on 24/7. Paired with MiniMax M2.1, it offers a responsive, cost‑effective AI coworker that automates repetitive information‑processing tasks, freeing users to focus on higher‑value work.