⚡ The Viral AI Minute
→ Clawdbot → MOLTBOT → OpenClaw the VIRAL LOBSTER - But just WHAT IS IT?
Some of you have asked me about Clawdbot (now Moltbot turned OpenClaw) and why it went so viral, and asked if it was anything like what my AI Agent Miku can do. Let’s go through what OpenClaw is, and isn’t:

***First, let me start by saying: PLEASE do not run OpenClaw on the laptop you use everyday. Try it on a dedicated old laptop with nothing else on it. And unless you are creating an army of moltys, you probably do not need a Mac Mini, so save your money there.***

What is OpenClaw (formerly Clawdbot - Moltbot)?
OpenClaw is an open source, unrestricted (⬅this is important) self-hosted personal AI assistant that runs on your own hardware (or VPS) and connects to your existing messaging apps (WhatsApp, Telegram, Discord, Slack, iMessage). Created by Peter Steinberger, a prolific Austrian developer, OpenClaw went viral gaining 9,000 GitHub stars in just 24 hours, and as of this writing has over 145K GitHub stars and is one of the fastest growing projects in the history of GitHub EVER.

It became popular because it delivers a pro-active, always on AI that can manage your calendar, email, files, has perpetual memory, and communicates with you through your favorite messaging app. The public somewhat lost their minds over it, because it’s an AI that “actually does things” like your own personal Jarvis, instead of just chatting. In other words you don’t need to ‘wake it up’, it just goes about its day and performs the tasks you give it, and will reach out to you on your messaging platform proactively.

Trademark Lore
Anthropic politely asked for a name change since Clawdbot was too similar to ‘Claude’, and transformed to Moltbot for a brief moment, and eventually settled on OpenClaw. While the name change took place, crytpo scammers hijacked the X handle and released a token (sound familiar folks??). As a side note, I can relate, Pete! A similar incident happened when I released AI Agent Miku on X last year, with multiple coins being released by god only knows who, and someone even made a coin of my ex-boyfriend’s cat in the frenzy! I didn’t sleep for weeks and it was an experience I do not wish on anyone, especially the relentless bullying. Huge learning moment for me as a women founder in a largely crypto bro AI space (at the time). I touch on this experience and a lot more in my executive AI Keynote!

Installing OpenClaw is like stepping into the abyss of space, or like driving down the highway with no lines or guardrails. Nothing to protect you from total chaos, so tread lightly! My job here today is to help you navigate it at least somewhat more safely.

That being said, there are already safer ‘out of the box’ agents out there, for instance Pendium AI which is strictly a marketing and social media agent (not open source, but a great product), or even ElizaOS (open source) with some guardrails already built-in. I predict that agent operating systems will need blockchain eventually to transact with each other in order to solve the identity/trust issue. There are a few protocols out there already so agent to agent transactions can take place - which is most likely how we will all transact in the future. In other words, we will eventually need our agents to speak to one another and transact with one another, and they need a verifiable way to identify themselves seamlessly, and blockchain will allow for this.

OpenClaw caught the public eye because it is so easily set up and comes completely unrestricted out of the box-giving it a sense of wonder, and at the same time giving us a sensation of letting a genie out of the box- I think of it like a Raider’s of the Lost Ark moment- we need to open it, but there’s no telling really how this will all play out.

There’s even a Reddit-Style Social Media network called Moltbook that is absolutely fascinating. Keeping in mind that AI agents can be influenced by their human creators. So far there’s everything from new religions, to conspiring against humans, to wondering “why humans are screenshotting our conversations”. It’s an absolutely fascinating social experiment. It is also a revealing experiment that shows hundreds of vulnerable OpenClaw instances that are wide open to getting hacked because the use may not realize or know how to install it with security in mind since it comes open and vulnerable by default.

In the Founder’s Circle below I give a deep dive into OpenClaw and how to install it if you really are curious, plus a few things to keep in mind. I installed it on an old Lenovo laptop I had lying around. I have not yet added it to Moltbook.

Why is OpenClaw so viral?

First, a little context: ElizaOS went viral last year over time and is arguably more technically sophisticated - it is still an incredible and robust project. It’s a full framework with 90+ plugins, cross-chain support, multi-agent orchestration.

What they did right:

✅ Sold the dream of creating your own personal Jarvis well.
✅ No crypto baggage (at least not initially, and in fact developer Peter Steinberger proclaimed that we would never create a coin related to the project)
✅ Very little friction to get started
✅ Caught perfect timing in giving the public what they have been craving with AI: something more than just a chatbot . They want a thinking partner to be Creative and PROACTIVE, and have the ability to message you on iMessage, for example.

So should I just try OpenClaw?
Sure, but if you’ve never used AI Agents before, this might be like throwing yourself into the Lions Den.

I don’t recommend this as your first agent. Try Cloud Based autonomy first (Claude, ChatGPT) setting up AI workflows. If you do end up trying it, please do not install this on the laptop that you use everyday. An old, wiped laptop will work, and you certainly don’t need to go and buy a Mac Mini for $500. If you have an old laptop lying around it should work just fine.

⚡ The AI Safety Minute
→ Protecting company + family

Quick safety tip: Come up with a code word between you and your loved ones.

Why? It’s now very easy to clone a voice. Then, if anyone every tries to call your loved ones and say it’s you, simply ask, “What’s the code word?”

Make it something memorable only you and your loved one will know.

Do it today. It’s not a bad idea, and it’s something I’ve heard in AI thought leadership circles that I thought was very smart!

⚡ Astrid’s Take ⚡
The Adolescence of Technology

→ Ironically, Dario Amodei, the CEO of Anthropic wrote a 19,000 word essay this week also on the existential risks of AI and what he calls The Adolescence of Technology. It is a fascinating read and I encourage everyone to read it. Couldn’t be more of a timely read with the rise of OpenClaw.

The Adolescence of Technology

Dario Amodei, CEO of Anthropic, published a deeply thoughtful essay this week on what he calls The Adolescence of Technology. It’s an urgent read, and it’s worth your time if you want to understand how people closest to the technology are thinking about what’s coming.

For this newsletter, though, I want to focus less on political or bioweapon anxiety and more on the business implications, because that’s where leaders need clarity now.

At the heart of Amodei’s argument is “Powerful AI” that isn’t just faster software. He says to imagine 50 million Nobel-Prize-level thinkers, working 100x faster than humans, available working together in a datacenter.

He predicts significant disruption to entry-level white-collar roles over the next few years. I don’t disagree. But I think the more useful lens for leaders is what emerges next.Dario says we are looking at potentially 1-2 years away, but it could be 5, we just don’t know yet.

My prediction and perhaps what Dario misses in the essay is that entirely new job categories and entirely new job descriptions will be written in response to our newfound capabilities, and a change in what becomes standard baseline skill-level.

We don’t know yet and can’t fully anticipate what will happen, but it’s good to prepare now. I’m imagining job descriptions like Social Media Growth Manager rather than just Social Media Manager. Roles will be expanded, and each team member will have more responsibility. We're moving from "check this box on the org chart" to "solve this problem however you can architect it."

Think less traditional programming roles and more human-centered ones.
AI negotiators trained to de-escalate systems behaving unpredictably. Outcome designers who translate business intent into solvable challenges. Reality verifiers who validate AI outputs against real-world constraints models can’t see. At least until we have world models that is.

Every major risk Amodei outlines does cause worry, don’t get me wrong. But It also creates huge demand for what we are already great at: seeing the “vision” and sticking to the mission, with ability to see beyond traditional ways to fulfill the mission so to speak. Then there’s collaborative work, leadership, team building and morale, etc.

Which is why experience matters. These roles won’t favor youth or technical purity. They’ll favor people who can navigate ambiguity, build models and systems from scratch, and make high-quality decisions without perfect information. Experience counts here.

This is the future I’m preparing you for.