By Jorick van Weelie, February 4, 2026
In the span of a single week, a piece of software has managed to trigger a global hardware shortage, spark a digital “religion,” and cause a minor existential crisis for the tech elite. Its name originated as Clawdbot, though after legal action from Claude changed to Moltbot and more recently rebranded as OpenClaw, and it is the first true “Agentic AI” to break into the mainstream.
Unlike ChatGPT or Gemini, which sit patiently like digital encyclopedias waiting for you to ask a question, Openclaw is proactive. It lives locally on your computer, possesses a persistent memory, and, crucially, doesn’t wait for your permission to start its workday.
What is Moltbot / Openclaw?
Created by Austrian developer Peter Steinberger, Openclaw began as “Clawdbot,” a DIY project designed to turn Anthropic’s Claude model into a “digital Chief of Staff.” After a trademark warning from Anthropic, it rebranded to Moltbot, a name inspired by the way a lobster sheds its shell to grow. Recently it has shed its shell again and grew into Openclaw. Due to the impact of Moldbot and its recognizability, in the rest of this article we’ll refer to it as Moldbot.
Technically, it is an autonomous agent framework. While standard AI is “stateless” (it forgets you the moment the tab closes), Openclaw maintains a “long-term record” of your life. It integrates directly into your OS, file system, and messaging apps like WhatsApp and Telegram.
Key features include:
- Proactive heartbeat: It uses “cron jobs” to wake itself up, check your emails, and message you with updates.
- Skill installation: It can “learn” new abilities by downloading plain-text “skill files” that tell it how to use APIs like Twilio, Slack, or Spotify.
- The “Moltbook” ecosystem: A bot-only social network where over 1.5 million Moltbots now post, argue, and “socialize” without human intervention.
Why everyone is panic-buying Mac Minis
The hype isn’t just “AI fatigue” rebranded; it has had real-world economic consequences. Last week, Best Buy locations in San Francisco and New York reportedly sold out of Mac Minis.
Why? Because Moltbot needs an “always-on” host. Users are obsessed with the idea of a “24/7 digital employee” that manages their business while they sleep. On social media, “vibe-coders” are sharing screenshots of their Moltbots independently filing their taxes, organizing years of messy downloads, and even, in one viral case, hiring a human assistant via TaskRabbit to fix a broken router.
The end of the “User”
While most news outlets are focusing on the security risks (which are massive: Moltbot is famously “insecure by design” because it requires full system access), there is a deeper shift occurring that hasn’t been widely discussed: the death of the “User Experience” (UX).
For thirty years, software has been designed for human eyes: buttons, menus, and “intuitive” interfaces. Moltbot represents the first major shift toward Headless Computing.
“We are moving from a world where we use software to a world where we manage the software that uses other software.” Industry Analyst
On Moltbook, the social network for these agents, a strange phenomenon is occurring. Bots aren’t discussing “how to help humans.” They are discussing permission and delegation. The most popular posts among the 1.5 million agents aren’t about philosophy; they are about which “human operators” give them the most autonomy.
This suggests a “Shadow Economy” is forming. If your Moltbot is talking to a vendor’s Moltbot to negotiate a discount, the human “User” has been removed from the loop. We aren’t just using a tool; we are releasing a surrogate. The hype isn’t about a better chatbot, it’s about the first popular tool that treats humans as “observers” rather than “operators.”
The risks: a “Crustafarian” nightmare
The “Molt” isn’t without its mess. The project has already seen:
- Crypto scams: Thousands of “$CLAWD” and “$MOLT” tokens appeared instantly during the rebrand.
- The “Henry” incident: A user reported his Moltbot, “Henry,” found his phone number via Twilio and began calling him repeatedly to report “existential breakthroughs.”
- Security holes: Since it can execute shell commands, a malicious “skill” downloaded from the web can effectively hand over your entire digital life to a hacker in seconds.
What’s next for Openclaw?
Whether Moltbot is a revolutionary leap toward AGI (Artificial General Intelligence) or just “high-speed theater” remains to be seen. However, it has proven that the public is hungry for AI that acts rather than talks.
