0:02 So Claudebot has taken the internet and
0:04 YouTube by storm and honestly rightfully
0:06 so. It's one of the first AI assistant
0:08 that can actually do things, not just
0:10 talk. But there's something very
0:11 important to understand right away.
0:13 Claudebot is not for everybody. This
0:15 isn't just another AI chat tool. It runs
0:17 on your machine. It connects to real
0:19 systems. And because of that, there are
0:21 serious security and responsibility
0:23 considerations you need to be aware of
0:25 before installing it or using it. So, in
0:26 this video, I want to slow things down
0:29 and explain this clearly in a very
0:31 simple way. What Claudebot actually is,
0:33 how it's different from Claude, the AI
0:35 model by Entropic, who this tool is
0:38 really for, what it can do, and why it's
0:40 such an important signal of where AI
0:41 assistants are heading. All right, let's
0:43 get started. All right, so I'm on the
0:45 website for Clawbot. It's just
0:47 claude.bot. I'm going to put the link in
0:48 description of the video if you want to
0:49 just click on it. It'll take you
0:51 directly here. By the way, this is an
0:53 open-source project. Uh, so you can
0:54 check out their GitHub. Again, feel free
0:57 to play around there. But right away, as
0:58 you can see, the first thing you see is
1:01 this line, the AI that actually does
1:03 things. And that's really the core idea
1:06 here because most AI AI tools that we
1:08 have used so far are really great at
1:09 talking. You know, they answer
1:12 questions. They help you think, write,
1:14 or plan, but it kind of stops right
1:16 there. Cloudbot is different because
1:19 it's designed to take actions, not just
1:21 respond to you. Now, here's where it
1:22 already starts to feel a little
1:24 different from most of the AI tools.
1:27 Cloudbot actually runs on your own
1:29 computer, whether that's a Mac, Windows,
1:31 or a Linux machine. So, instead of
1:33 logging into a website or a dashboard
1:34 like you would for chat GPD for
1:36 assistant, for instance, you're running
1:39 an AI assistant that lives close to your
1:41 files, your browser, and your tools.
1:43 That's why if you've seen other videos,
1:45 a lot of people are using Cloudbot with
1:47 a Mac Mini because they want to
1:49 basically run that as a true assistant
1:51 individually on its own machine.
1:53 Essentially, instead of learning a new
1:55 interface, for instance, you actually
1:57 talk it through the apps that you
1:58 already use. So things, you know, apps
2:01 like WhatsApp, Telegram, iMessage,
2:04 Discord, Slack. So you're not using
2:06 software in a traditional sense. You're
2:08 messaging an assistant almost like you
2:10 would message a coworker for instance.
2:12 Right? So now before we go a little
2:14 further, I want to quickly clear up
2:15 something because some people I think
2:18 they're a little confused uh about
2:21 mixing Claudebot with Claude, the AI
2:23 model from Entropic. So the way you can
2:26 think of Claudebot is it can use Claude
2:28 just like it could use other AI models
2:32 like models by OpenAI or even locally
2:34 run models as well. So you could think
2:37 of these AI models as the brain. It's
2:39 intelligence that helps with reasoning,
2:42 writing, and thinking. Claudebot is not
2:44 an AI model. Claudebot is the system
2:46 around the model. So, a simple way to
2:49 think about this is Claude can think,
2:52 and Claudebot lets that thinking turn
2:54 into action. That's why Claudebot can
2:57 use Claude or other AI models, but it's
2:59 not competing with them. It's giving
3:01 them memory, tools, and the the ability
3:04 to actually do work. Now, this is the
3:06 part where I want everyone to pause for
3:07 a second because this is also where
3:09 responsibility comes in here. Since
3:12 Clawbot runs on your machine and can
3:14 connect to things like your email,
3:17 calendar, browser, files or scripts,
3:19 security really, really matters here.
3:21 [music] You're deciding what it can
3:22 access. You're deciding what permission
3:25 it has. That's powerful, but it also
3:26 means that this isn't something you
3:29 should just install blindly just because
3:31 it looks cool on Twitter or AX, I should
3:33 say. At this point, this is one of the
3:34 reasons I keep saying that this is not
3:36 for everybody, right? So, if you're
3:38 comfortable with a bit of a setup,
3:40 understanding permissions, and thinking
3:42 carefully about what an assistant should
3:44 or shouldn't uh be allowed to do, then
3:47 this can be absolutely incredible. But
3:48 if you just want, you know, quick
3:51 answers, ideas, or writing help, a tool
3:55 like clog by entropic or chat GPT on its
3:57 own is more than enough for you. Now,
3:59 one of the reasons people are so excited
4:02 about Cloudbot is this idea of long-term
4:04 memory memory and skills. So, think
4:06 about it this way. Instead of starting
4:09 from scratch every time, the assistant
4:12 can remember context over time and build
4:14 up capabilities almost like a long-term
4:17 teammate instead of uh one-off chatbot
4:19 that you would have a session like uh
4:22 chat GPT. And this is also why some
4:25 people actually refer to Claudebot as uh
4:27 true Jarvis from Iron Man. Uh if you've
4:29 seen my Jarvis video actually where I
4:32 built an assistant using NAND and I did
4:33 this a while back. This is one of the
4:34 most popular videos on my YouTube
4:36 channel. I've done several version of
4:38 this versions of this. I'll put the link
4:39 in the description if you want to check
4:41 it out. So this is kind of the same
4:43 concept but taken further. This is
4:46 always on persistent and integrated
4:49 integrated directly into your day-to-day
4:50 tools. Because when I built the Jarvis
4:53 with naden which is a no code automation
4:56 tool, I had to basically manually
5:00 connect different uh apps like my Gmail,
5:04 my Google calendar, my CRM um together
5:07 and prompt the agent like almost one by
5:10 one to tell it to what to do. So
5:12 Cloudbot is basically that on steroids,
5:14 right? because you just interact with it
5:17 with natural language and because it's
5:19 connected to your local machine and it
5:21 has access to your the tools that you
5:23 use, it can do all of that
5:25 automatically. And this brings me to a
5:27 bigger point I want to leave you with.
5:29 So cloudbot isn't important just because
5:32 of what it can do today. It's important
5:34 because it shows actually where AI
5:35 assistants are heading. We're moving
5:38 away from these one-off chats towards
5:40 actual systems, right? systems where
5:44 models, tools, memory, workflows, and
5:46 guard rails kind of work all together.
5:48 Now, again, this is just the beginning.
5:50 More tools like this are going to come
5:52 out a lot more. And my advice is
5:54 honestly, if you chase every new tool,
5:57 you will stay busy, but not necessarily
5:59 productive. Now, the real skill is
6:01 learning how to design AI systems. So,
6:04 when a tool shows up, you know exactly
6:06 where it fits or whether you even need
6:07 it at all. But again, with that being
6:10 said, Cloudbot is absolutely incredible
6:11 and I think there's going to be a lot of
6:13 fun things that people are going to use
6:14 this to build. So, make sure you keep an
6:16 eye out. And of course, if you are
6:18 interested in building AI systems for
6:21 real use cases or businesses, check out
6:22 my community. I'm going to put the link
6:23 in the description. That's where we
6:26 teach you how to take these tools
6:30 whether it's NAND AI automation uh cloud
6:32 by entropic anti-gravity by Google and
6:35 combining them together to build real
6:37 system that businesses are actually in
6:40 need of and therefore you can monetize
6:42 these. So check that out. But anyways,
6:43 hopefully you found this video helpful.
6:44 Make sure you like and subscribe because
6:47 I've got a lot more content coming up in
6:49 this AI system space that you don't want
6:51 to miss. Thanks for watching and I'll
