ClawdBot is on everyone’s lips. Or at least that’s how it seems if you look at Twitter. I see a lot of hype, a lot of people installing it, but I also see people who have no idea what it really is.
So today I’m going to explain to you exactly what it is, how it works, what dangers it has, and why it is not for everyone. And watch out, I’m preparing another video on how to install it totally free.
What is ClawdBot?
ClawdBot, which is now called Moltbot because the Claude friends hinted at it and the developer (Nick) changed it without problem, is an open source project. You can see all its code.
For you to understand easily: it acts as an artificial intelligence agent, an assistant. Many people call it “what Siri or Alexa would want to be”. And they are right, but watch out, there is an important nuance: ClawdBot has a very technical background.
It is not like buying an Alexa and plugging it in. You need to install it on a computer or on a VPS, it doesn’t matter if it’s local or in the cloud, but you need a computer.
Why is it different from ChatGPT?
Here comes the interesting part. With ClawdBot we can configure any artificial intelligence model. Any. We can use open source models that run locally, we can use powerful models like Claude’s, we can configure whatever model we want.
It is not like ChatGPT that only uses GPT, nor like Claude that only uses Claude. ClawdBot is versatile.
And once the model is configured, the next thing is to add a personality. Yes, a personality. You can make it speak to you formally, tease you, whatever you want. ChatGPT has something similar with custom GPTs, but it is nowhere near as configurable as this.
The Skills: here is the magic
The next super important point is versatility. ClawdBot has programmed skills to perform actions. And when I say actions, I mean everything:
- Skills to make phone calls
- Skills to manage your Notion
- Skills to tweet
- Skills to manage your email
- There are skills for everything!
And the advantage is that ClawdBot is smart enough so that, through a chat that you configure (many people use Telegram or WhatsApp), you can write to it with natural language and it knows what it has to do.
This is what is driving people crazy.
You take your ClawdBot, configure the email, calendar, your notes, whatever you want, and then ask it for things. For example:
- There are people who connect it to their watch and say: “Every morning at this time tell me how my vital signs are, if I slept well and what I can face today.”
- Or you can tell it: “When I send you an audio talking about X topic, create a document in Notion and help me develop a YouTube script.”
In the end, with natural language you are telling it to do anything: scheduled tasks, crons, whatever. And it also has memory. You can tell it “remember that I want you to talk to me like this” or “next time we talk remind me of X thing”, and it does it.
It’s like n8n, but talking
What I am seeing is that with ClawdBot we can do things very similar to what we did with n8n, but in a super simple way. You don’t have to complicate yourself creating workflows, you simply talk to it, synchronize the platforms and that’s it. It works.
Think of it like in the movies Her or Iron Man with Jarvis. In the end it is a bit like that. It has a very big difference with ChatGPT: the difference of being able to DO things. Not just answer you, but execute actions.
It is a bit where AGI, general artificial intelligence, is focused. Obviously ClawdBot is not an AGI, but since it has so many tools to connect to other places, it can help you practically in everything. The limits are your imagination.
Watch out: the dangers no one tells you
But ClawdBot has disadvantages, and this is important. It is in beta and, as I said, it is very technical.
Many people are installing it without knowing exactly what they are doing. And then surprises arrive:
- Unexpected costs
- Improper access
- Actions you didn’t want it to take but it has done because you don’t have enough control
Without going any further, I was tinkering so that it would recognize me when I spoke to it by audio, and it changed a configuration itself that made everything break. I had to connect to the server to fix it manually.
It’s not a problem for me because I’m a programmer, but someone non-technical wouldn’t know what to do. They would panic and the AI could no longer help them because it was broken.
Be very careful: it is not simple if you do not have prior knowledge.
Costs: another important issue
Also the token expenses are tremendous. I am currently using the ChatGPT subscription and well, thank goodness it is generous. But there are crazy people who are using the Claude API and are spending a lot of euros a month.
You have to know well which model to use and when.
Conclusion
In the end, ClawdBot is a super practical tool but one that must be used with caution. It is an assistant to which we can configure anything simply. There is no more, that is the whole explanation.
I don’t want to sell you more smoke. It is not blessed glory either, but it is something that impresses and opens the door to many things. I myself am preparing little things that I hope will serve me in the future.
And I remind you: I am going to make a tutorial on how to install it. But watch out, I am going to tell you what no one tells you: how to do it safely, security tips that are super important and that almost no one is taking into account.
So if you are interested in that tutorial, follow me because I upload it shortly.
O que você achou?
Deixe sua opinião, pergunta ou sugestão. Os comentários são sincronizados com GitHub Discussions .