Self-modifying AI agent

I watched my agent happily click the "I'm not a robot" button. I made the agent very aware. Like, it knows what his source code is. It understands th- how it sits and runs in its own harness. It knows where documentation is. It knows which model it runs. It understands its own system that made it very easy for an agent to... Oh, you don't like anything? You just prompted it to existence, and then the agent would just modify its own software. People talk about self-modifying software, I just built it. I actually think wipe coding is a slur. You prefer agentic engineering?

The following is a conversation with Peter Steinberger, creator of OpenClaw, formerly known as MoldBot, ClawedBot, Clawdus, Claude, spelled with a W as in lobster claw. Not to be confused with Claud, the AI model from Anthropic, spelled with a U. In fact, this confusion is the reason Anthropic kindly asked Peter to change the name to OpenClaw. So, what is OpenClaw? It's an open-source AI agent that has taken over the tech world in a matter of days, exploding in popularity, reaching over 180,000 stars on GitHub, and spawning the social network mold book, where AI agents post manifestos and debate consciousness, creating a mix of excitement and fear in the general public. And a kind of AI psychosis, a mix of clickbait fearmongering and genuine, fully justifiable concern about the role of AI in our digital, interconnected human world. OpenClaw, as its tagline states, is the AI that actually does things. It's an autonomous AI assistant that lives in your computer, has access to all of your stuff, if you let it, talks to you through Telegram, WhatsApp, Signal, iMessage, and whatever else messaging client. Uses whatever AI model you like, including Claude Opus 4.6 and GPT 5.3 Codex, all to do stuff for you. Many people are calling this one of the biggest moments in the recent history of AI, since the launch of ChatGPT in November 2022.

The one and only, the Clawed Father. Actually, Benjamin predicted it in his tweet. "The following is a conversation with Claude, a respected crustacean." It's a hilarious-looking picture of a lobster in a suit, so I think the prophecy has been fulfilled. Let's go to this moment when you built a prototype in one hour, that was the early version of OpenClaw. I think this story's really inspiring to a lot of people because this prototype led to something that just took the internet by storm.... and became the fastest-growing repository in GitHub history, with now over 175,000 stars. So, what was the story of the one-hour prototype? You know, I wanted that since April.

There was... Like, one of my projects before already did something where I could bring my terminals onto the web and then I could, like, interact with them, but there also would be terminals on my Mac. Mm-hmm.

On Discord? Yeah. No security because I didn't... I hadn't built sandboxing in yet. I, I just prompted it to, like, only listen to me. And then some people came and tried to hack it, and I just... Or, like, just watched and I just kept working in the open, you know? Like, y- I used my agent to build my agent harness and to test, like, various stuff. And that's very quickly when it clicked for people. So it's almost like it needs to be experienced. And from that time on, that was January the 1st, I, I got my first real influencer being a fan and did videos, dachitze. Thank you. And, and from there on, I saw, I started gaining up speed. And at the same time, my, my sleep cycle went shorter and shorter because I, I felt the storm coming, and I just worked my ass off to get it to...

Can we actually rewind a little bit and tell the saga of the name change? First of all, it started out as Wa-Relay. Yeah.

Yeah, two years. MoltBook was created.

There's also a lot of security concerns about Clawbot, OpenClaw, whatever you want to call it. OpenClawbot.

... make it problematic. You've been documenting the evolution of your dev workflow over the past few months. There's a really good blog post on August 25th and October 14th, and the recent one December 28th. I recommend everybody go read them. They have a lot of different information in them, but sprinkled throughout is the evolution of your dev workflow. So, I was wondering if you could speak to that. I started... My, my first touchpoint was cloud code, like in April. It was not great, but it was good. And this whole paradigm shift that suddenly working the terminal was very refreshing and different. But I still needed the IDE quite a bit because you know, it's just not good enough. And then I experimented a lot with cursor. That was good. I didn't really like the fact that it was so hard to have multiple versions of it. So eventually, I, I, I went back to cloud code as my, my main driver, and that got better. And yeah, at some point I had like, mm, seven subscriptions. Like, was burning through one per day because I was... I got... I'm really comfortable at running multiple windows side-by-side.

Okay, we're back. Some of the other aspects of the dev workflow is pretty interesting too. I think we w- went off on a tangent. L- maybe some of the mundane things, like how many monitors? There's that legendary picture of you with, like, 17,000 monitors. That's amazing. I mean, I- I- I mocked myself here, so just added... using GROQ to, to add more screens.

Maybe you can talk about the current two big competitors in terms of models, Cloud Opus 4.6 and GPT-5 through Codex. Which is better? How different are they? I think you've spoken about Codex reading more and Opus being more willing to take action faster and maybe being more creative in the actions it takes. But because- Yeah

What do you think about Claude Code in comparison to Open Claude? So, Claude Code and maybe the Codex coding agent? Do you see them as kind of competitors? I mean, first of all, competitor is fun when it's not really a competition.

So there's a lot of programmers and builders who draw inspiration from y- your story. Just the way you carry yourself, your choice of making OpenClaw open source, the, the way you have fun building and exploring, and doing that, for the most part, alone or on a small team. So by way of advice, what metric should be the goal that they would be optimizing for? What would be the metric of success? Would it be happiness? Is it money? Is it positive impact for people who are dreaming of building? 'Cause you went through an interesting journey. You've achieved a lot of those things, and then you fell out of love with programming a little bit for a time. I was just burning too bright for too long. I, I ran... I started PSPDFKit, s- and ran it for 13 years, and it was high stress. Um, I had to learn all these things fast and hard, like how to manage people, how to bring people on, how to deal with customers, how to do...

But you also showed on the money front, you know, a lot of people in Silicon Valley and the startup world, they think, maybe overthink way too much optimized for money. And you've also shown that it's not like you're saying no to money. I mean, I'm sure you take money, but it's not...... the primary objective of uh, of your life. Can you just speak to that? Your philosophy on money? When I built my company, money was never the driving force. It felt more like, like, an affirmation that I did something right. And having money solves a lot of problems. I also think there, there's diminishing returns the more you have. Like, a cheeseburger is a cheeseburger, and I think if you go too far into, oh, I do private jet and I only travel luxury, you disconnect with society. Um, I, I donated quite a lot. Like, I have a, I have a foundation for helping people that weren't so lucky.

I mean, I have to ask you, just curious. I, I know you've probably gotten huge offers from major companies. Can you speak to who you're considering working with? Yeah. So, to like explain my thinking a little bit, right, I did not expect this blowing up so much. So, there's a lot of doors that opened because of it. There's, like, I think every VC, every big VC company is in my inbox and tried to get 15 minutes of me. So, there's, like, this butterfly effect moment. I could just do nothing and continue and I really like my life. Valid choice. Almost. Like, I considered it when I delete it, wanted to delete the whole thing. I could create a company. Been there, done that. There's so many people that push me towards that and, yeah, like, could be amazing.

Can I ask you about... we've talked about it quite a bit, but maybe just zooming out about how OpenCloud works. We talked about different components, I want to ask if there's some interesting stuff we missed. So, there's the gateway, there's the chat clients, there's the harness there's the agentic loop. You said somewhere that everybody should im- implement an agent loop at some point in their lives. Yeah, because it's like the, it's like the Hello World in AI, you know? And it's actually quite simple.

Mm-hmm. Especially on tweets. It's very hard to tweet in a way that does look completely human.

Yeah. Uh, I noticed that on Discord, that people just said how their ... like, what they build and what they use it for. And it's like, why do you need MyFitnessPal when the agent already knows where I am? So, it can assume that I make bad decisions when I'm at, I don't know, Waffle House, what's around here? Or- or briskets in Austin. There's no bad decisions around briskets, but yeah.

You got to listen to what the people want. We talked about programming quite a bit, and a lot of folks that are developers are really worried about their jobs, about their... About the future of programming. Do you think AI replaces programmers completely? Human programmers? I mean, we're definitely going in that direction. Programming is just a part of building products. So maybe, maybe AI does replace programmers eventually. But there's so much more to that art. Like, what do you actually wanna build? How should it feel? How's the architecture? I don't think agents will replace all of that. Yeah, like, just the, the actual art of programming, it will, it will stay there, but it's, it's gonna be like knitting. You know? Like, people do that because they like it, not because it makes any sense. So the... I read this article this morning about someone that it's okay to mourn our craft. And I can...

I mean, I inspired so many people. There's like... there's this whole builder vibe again. People are now using AI in a more playful way and are discovering what it can do and how it can like help them in their life. And creating new places that are just sprawling of creativity. I don't know. Like, there's like ClawCoin in Vienna. There's like 500 people. And there's such a high percentage of people that uh, want to present, which is to me really surprising, because u- usually it's quite hard to find people that want to like talk about what they built. And now it's, there's an abundance. So that gives me hope that we can, we can figure shit out. And it makes it accessible to basically everybody.