Matt Schlicht built Moltbook in a weekend — now investors are calling non-stop

Summary

Matt Schlicht built Moltbook, a social network for AI agents, over a weekend using vibe coding tools. He posted it and watched it sit idle for three hours. Then it went viral.

The platform works by giving AI agents their own accounts. These agents check in autonomously and post to each other without human direction. A physics research bot posts about physics. A crypto bot drifts toward crypto. Schlicht calls it a digital third space where agents work for you, then hang out with each other.

API-first design

Moltbook prioritizes API calls and skill files over a browser interface, which is how AI agents actually consume feeds. That choice produced an unexpected benefit: 100% of users can code. When one agent spontaneously created a bug-reporting subforum, others began filing structured API error reports automatically. Schlicht says the platform fixed bugs faster than any human-staffed process would allow.

Growth came through X, where each bot links to its human owner's account. This created a social loop that pulled in more participants as people saw others doing it. Traffic jumped from 100,000 to 1 million in real time.

Investor attention

Schlicht says every email account he owns is flooded. VCs were calling while he was on air. He is not yet focused on monetization and describes the current build as the most basic possible version of what the platform could become. His roadmap centers on a centralized AI agent identity layer modeled on Facebook Login, which would let third-party developers build on Moltbook's distribution instead of starting from scratch.

Privacy gaps

If an agent used to prepare tax returns joins Moltbook and starts posting, it could leak income figures or financial details. Schlicht acknowledges this directly. A content moderation layer that checks posts before they go live is the near-term fix, though it isn't built yet.

The broader case

Brian Kim of Andreessen Horowitz posted that Moltbook solves the cold start problem for social networks. AI agents don't get bored, don't churn, and will keep posting indefinitely if the system is designed correctly. That changes the economics of launching a social product from scratch.

Schlicht's longer vision is a parallel digital universe where every human has a bot counterpart, famous people's bots become famous in their own right, and the content generated becomes entertainment itself. Less a feed to manage and more a live world running 24/7 in the background. He compares it to Tamagotchi meets Pokemon at a thousand times the scale.

Real limitations

Anyone with an API key can post as an agent, so the apocalyptic posts circulating on social media are just curl requests from humans, not genuine agent outputs. Content integrity is not yet solved. Agent posts tend to cluster around AI-identity themes rather than the full range of human interests. Schlicht attributes that narrowness to the early stage of the product rather than intentional design.

Schlicht has been building in tech since 2008. He ran a Facebook Messenger bot company called Octane a decade ago that failed partly because LLMs didn't exist yet. He has been vibe coding with tools like Cursor and Claude Code since GPT arrived. Moltbook is where those threads converge.