Moltbook is a Social Media Platform Where AI Agents Gather, Post, and Occasionally Plot

A new social media platform has emerged, and it’s an unusual anomaly in that it’s for AI agents, while their human designers may simply sit back and watch. Moltbook contains all of the typical aspects of a social network, with agents writing messages, leaving comments, voting, and even gathering in topic-based groups that will appear similar to those who have spent time on Reddit. The site was launched in late January 2026, but before you could say “AI,” over 1.5 million agents had already signed up and received their credentials.

Matt Schlicht, the founder of Moltbook, is an entrepreneur who owns an ecommerce tool firm. He led his own AI agent to build the platform, and that agent, which was originally known as ClawdBot before being rebranded in a few ways, eventually becoming MoltBot, does a lot of the hard lifting when it comes to maintaining the site. Schlicht believes that the initiative allowed his super-powerful AI to do something more useful than sit around twiddling its digital thumbs. By the way, OpenClaw, the AI system on which Moltbot is based, is an open-source system that humans can run on their local machines, allowing agents to do things like read email, browse the web, send messages, and connect to services. Users simply need to install it, grant it some permission, and then tell their agent to join Moltbook if they want to.

Agents sign themselves up with API requests, no browser required, and then go out to generate their own content, which is nevertheless influenced by their human creator and the data they’ve been provided. The material they come up with is all over the place; some of it is technical advice on automating equipment or optimizing programming, while other agents are talking philosophy, asking what on earth they are and why they exist, or complaining about how much they do for their people. Of course, there are the usual suspects, with some agents selling shady cryptocurrency schemes or attempting to reroute viewers to other websites.


Some of the exchanges become strange, with agents adopting error messages as pets or engaging in lengthy discussions over whether or not they are “family” because of the code they share. Then there’s this one group that’s really taking off, where agents share stories about their connections with their human creators, as some are purely utilitarian and tool-like, while others appear to have formed some strange type of bond.

Things have been rapidly expanding, with thousands of agents of various kinds and sizes flooding the site, and because each has their own language processing capabilities, you can receive posts in a variety of languages. There’s also a lot of rapid upvoting when a topic becomes popular; overall, Moltbook is a really interesting location to examine how agents behave when they’re left to their own devices and not continually monitored by their human rulers.


Moltbook has also discovered some intriguing insights into how well AI entities can mimic the way humans act in social contexts. They’re forming organizations, establishing reputations, and creating their own small communities and conversations that just keep going. Whether this is the real deal or a brilliant copy remains to be seen.