OpenClaw, a powerful new agent assistant, has a thing for guacamole.

This is one of the many things I discovered while using viral artificial intelligence bot as my personal assistant this past week.

Formerly known as both Clawdbot and MoltbotOpenClaw has recently become a Silicon Valley darling, charming AI enthusiasts and investors eager to take a bleeding edge or profit from it. The highly capable, web-savvy AI bot is even inspiring own AI-only (or mostly) social network.

As the writer of WIRED’s AI Lab newsletter, I thought I should do it and try using OpenClaw myself. I have a bot that monitors incoming emails and other messages, digs up interesting research, orders groceries, and even negotiates deals for me.

For brave (or perhaps reckless) early adopters, OpenClaw seems like a legitimate glimpse into the future. But any sense of wonder is accompanied by a bit of panic as the AI ​​agent romps through emails and file systems, uses a credit card, and sometimes even turns on its human user (although in my case, this part of the face is my fault).

How I Set It Up

OpenClaw is designed to live on a home computer that is always on. I configured OpenClaw to run on a PC running Linux, to access the Anthropic model Claude Opus, and to communicate with me on Telegram.

Installing OpenClaw is simple, but configuring it and maintaining it can be a headache. You need to give the bot an AI backend by creating an API key for Claude, GPT, or Gemini, which you paste into the bot’s config files. In order for OpenClaw to use Telegram, I also had to first MAKE a new Telegram bot, then provide OpenClaw with the bot’s credentials.

For OpenClaw to be truly useful, you need to connect it with other software tools. I created a Brave Browser Search API account so that OpenClaw can search the web. I also configured it so that it can access the Chrome browser through an extension. And, God help me, I gave it access to the email, Slack, and Discord servers.

When it’s all done, I can talk to OpenClaw from anywhere and tell it how to use my computer. In the beginning, OpenClaw asked me some personal questions and let me choose its personality. (The options reflect the project’s anarchic vibe; my bot, called Molty, likes to call itself a “chaos gremlin.”) The resulting persona is very different from Siri or ChatGPT, and it’s one of the secrets of OpenClaw’s popularity.

Web Research

One of the first things I asked Molty to do was send me a daily roundup of interesting AI and robotics research papers from arXiv, a platform where researchers upload their work.

I used to spend a few afternoons vibe-coding websites (www.arxivslurper.com and www.robotalert.xyz) to find the arXiv. It’s amazing (albeit a little demoralizing) to see OpenClaw immediately automate all the same browsing and analysis work required. The roles it chooses are pretty good, but with more instruction I think it could be better. This kind of web search and monitoring is definitely helpful, and I thought I’d use OpenClaw for it.

IT support

OpenClaw also has an amazing, almost terrifying ability to fix technical issues on your machine.

This is not surprising, because it is designed to use a border model that makes it possible to write and debug code and use the command line easily. However, it’s scary when OpenClaw just reconfigures its own settings to load a new AI model or debug a browser problem on the fly.