# OpenAI Just Bought OpenClaw!!

## Метаданные

- **Канал:** Sam Witteveen
- **YouTube:** https://www.youtube.com/watch?v=cQeZY0FRMFE

## Содержание

### [0:00](https://www.youtube.com/watch?v=cQeZY0FRMFE) Segment 1 (00:00 - 05:00)

Okay, so forget chat GBT as you know it. The lobster has handed Sam Molman the keys to the world's second key AI agent. And I'm telling you right now, the chatbot era of LLMs has just died. So in this video, I'm going to break it down. We're going to look at some of the things that have led up to this and also what this means for agents going forward as a whole. So, originally today's video was supposed to talk about how you can actually set up Open Claw on some VPS systems and how you can do some various tasks with it. Well, clearly this has been surpassed by the fact that OpenClaw has now been kind of acquired by Open AI. Now, no one's calling it an acquisition. There's no talk yet that I've seen at least of how much money was paid for this, but we do have Peter Steinberger announcing that he's joining OpenAI to bring agents to everyone and that Open Claw is going to become a foundation, open, independent, and is just getting started. Now, the words foundation, open, and independent are perhaps not the best words to use when you talk about something with Open AI, considering they're currently getting sued for taking their open foundation and turning it into something that totally is not independent. But if we look at the announcement that Peter Steinberger has put up on his own blog, and by the way, there's some really nice posts that he has actually up there about how to get the most out of coding with Loom. But the post here is really focused on sort of why he decided to move on to open AI. And you got to think that this has basically come on the back of lots of VCs offering him probably 100 million plus for his first seed round if he were to turn this into a company. But I can totally get the fact that he's done that before is not something that he really wants to repeat. And it does seem to me that he's very genuine and actually not driven by trying to get the most amount of money. So the talk around town was that he was in negotiations with both open AAI and Meta. And we know that Meta themselves have been really big on sort of acquiring things around agents. Not that long ago they basically acquired Manis AI which is a full agent system. They also acquired Limitless AI, which is basically a device that you wear, which records all of your conversations and everything you hear and works really well at getting your sort of overall life context into a format that you can then put into an LLM. So, unless you've been under a rock for the last few months, you've probably seen Open Claw. And this project really did take the internet by storm. It sort of promised the holy grail of an AI agent that doesn't just talk to you, but actually does things for you. It browses, it clicks, it executes. And unfortunately for a lot of people, it also could get hacked because of the way that people were using it where people were going out and buying Mac minis or setting it up basically with root access to things and no security. But regardless of that, it clearly has been a sort of wakeup signal for people around agents. Now, in the introduction, I said that this was the world's second key AI agent. And what I want to remind people of is that we've kind of seen some of these ideas before, maybe not as good, but the whole sort of auto GPT moment was something that we saw a couple years ago where again people basically put these things into a loop, got it going. Now, obviously back then the models weren't as good. There weren't really tools in the same way. Apart from spending really big API bills on things like GPT4, people didn't get into too much trouble, I think, with this. Open Claw though, which if you haven't been paying attention, is actually the third name for this. This started out as Claudebot, changed to Maltbot, then became OpenClaw. It does have access to a lot of useful things, meaning it's got access to lots of tools. It's got sandboxes where it can write and execute its own code. It's got the ability to have a persistent memory. So, a lot of the things that people have been talking about with deep agents and some of the other things like that over the past 6 months, these were all sort of baked into this. On top of that, it was very easy to integrate things like Telegram or WhatsApp or Discord so that you had a way to be able to communicate with this agent wherever it was running. And I think that's one of the key things that we've seen this be such a big sort of deal. Now, what happens to this going forward is going to be very interesting to see. So basically Sam Olman has put up a post where he explicitly states that Steinberger is joining to drive the next generation of personal agents. So even open AI which has had multiple attempts at trying to do you know sort of agentic things to be a hit whether that goes back to their

### [5:00](https://www.youtube.com/watch?v=cQeZY0FRMFE&t=300s) Segment 2 (05:00 - 08:00)

agents API the agents SDK right through to their agentic browser atlas none of those things have become anywhere near the hit that open claw has become. So compared to openclaw, all of these things feel kind of clunky and limited. And you got to think by bringing Steinberger in, OpenAI didn't just hire a builder, they effectively acquired the most successful user interface for agents that's currently on the market. Now, at the same time that this has been a really big hit for Open AI, you've also got to wonder, has this been a really big missed opportunity for Anthropic? So this originally was called Claudebot spelled differently but it obviously was hinting at the using the model Claude and many people were using Claude to actually run this and Enthropic basically chased Steinberg away with a sort of cease and desist letter and gave him sort of like you know I think it was two or three days to actually change the name and change or any association between the project and Claude itself or face legal action. They even refused to let the old domains point to the new project. So instead of embracing the community that was building on top of their models, it probably basically used what many people have called sort of draconian tactics to actually distance themselves from this project. Now, you can understand that, you know, where they coming from with all the sort of security issues and things that were going on, but you also got to ask yourself in hindsight, would have they just been better to invite him into a free lunch in their canteen where they could actually talk about them actually taking over this project and bringing him to work there. So, I guess we'll never really know what the actual thinking there was, but it certainly does leave a lot of questions unanswered. And the biggest one of those questions has got to be is OpenClaw about to come closed claw. So even though Sam Wman has talked about that this is going to stay as an open-source project, they're going to have a lot of influence over where it goes. And I do suspect personally that probably what we're going to see happen is people just fork this code or build things that are similar and the community will take it from there. So I do think of some of the biggest lessons from this whole claudebot sojourn that we've been on for the past month or so has shown that it's not just hardcore techies that are interested in agents. Lots of people want to have agents that are running there for them. Lots of people are starting to wake up to just how powerful this technology is and what it can actually do when you've got a multi- aent system set up to be working for you sort of 24/7. And it does seem that the focus has officially shifted from what an AI can say to you to really what can an AI do for you. And I think while lots of people talked about 2025 as being the year of agents and stuff like that, we still were at very early stages back then, I still think we're reasonably early stages now. But we are moving toward a world where AI isn't just a tool that you use. It's a teammate that executes tasks on your behalf. So I definitely want to hear from all of you. Let me know in the comments what you think is actually going to happen to open claw. Has the claw closed. Has Sam Alolman killed this open-source project? And in 6 months from now, are we going to look back on this just like people look back on AutoGPT now as something that was kind of interesting at the time, but really was more a novelty rather than anything sort of useful for your day-to-day life? Anyway, let me know in the comments. Don't forget to hit like and subscribe if you found the breakground helpful and I will talk to you in the next one.

---
*Источник: https://ekstraktznaniy.ru/video/22382*