Master OpenClaw AI Agents

Vídeo em Inglês, publicado pelo canal Eduonix Learning Solutions.

About this content: The summary, key points and topics below are generated automatically from the video transcript. They reflect the content, opinions, claims and predictions presented in the video and do not represent a position, validation or fact-check by IA Dots.

Video summary

An introduction to Open Cloud, an open-source AI agent that operates locally or on a virtual private server to perform tasks, manage files, handle communication, and automate workflows using natural language instructions.

Key points

• Open Cloud is presented as an open-source AI agent rather than a passive chatbot, capable of planning, taking real actions, and iterating until a task is completed.
• The software runs locally on the user's machine or on a VPS, ensuring user data stays private while API calls go to chosen providers.
• The agent operates using a permission system, requiring user approval before executing actions like writing files, sending messages, or running commands.
• Core concepts include the agentic loop (plan, act, observe, and iterate), persistent memory stored as markdown files, and proactive behavior.
• Prerequisites for installation include Node.js, and users can choose between a local installation for learning or a VPS setup for 24/7 continuous operation.

Topics

AI AgentOpen SourceFile ManagementAutomationNode.jsVirtual Private ServerPermission System

Video transcript

Transcript from available captions; it may contain errors.

Welcome to Open Cloud for beginners. In this course, you are going to install, set up, and start using your own AI agent. Let's start with what Open Cloud actually is. So, Open Cloud is an open-source AI agent that runs on your computer. The keyword here is agent, not chatbot. A chatbot is like ChatGPT or Claude. Waits for you to type a question, gives you an answer, and that's it. An agent is different. You describe the task, and it plans, takes real actions, and checks the results. And also, it keeps going until the job is done. Open Cloud can manage your files, send messages through WhatsApp or Telegram, browse the web, check your calendar, draft emails, and connect to dozens of other apps. It does not really like uh talk about doing things, but it just like do the things. Now, you might ask the question, why it matters? Well, three things make Open Cloud stand out. First, it it's an open-source project. It's free. MIT license. 350,000 stars on GitHub. No subscription, no vendor lock-in. You own everything. Second, it runs locally on your machine. Your data stays on your computer. The only thing that leaves is the AI model calls, which go out to which provider you actually choose. And third, it's extendable, which means that there are thousands of community built-in skills on are on Cloud. Um you can connect almost any app and automate almost any workflow. Now, the story behind this is pretty much like amazing. Um Open Cloud started as a personal project called Cloud Cloud Bot. Cloud Cloud Bot in a way, but like it's very close to Cloud uh by a solo developer. It's got renamed to Mult Bot after some trademark concerns, again from like uh Cloud um specifically. So, it was called uh Mull bot and then like it was rebranded to Open Cloud and placed under an independent foundation with community governance. It went viral almost overnight. So, we had like 350,000 GitHub stars on the repo, 15,000 Discord members, and one of the fastest growing open-source project of the year. If you want to check it, uh you can go to Open Cloud and you can see like documentation and you can read about uh everything, especially on the contribution side. Next, uh let's see actually some examples of using Open Cloud. Now that you know what Open Cloud is, let's look at what it actually does. Um I want to walk through real examples so you can see the kind of task this agent can handle. First, we have file management. Here are three examples of how Open Cloud works with files on your computer. First, summarizing a document. You say, "Summarize proposal.pdf in three bullet points." And Open Cloud will just do it. It will extract the um key points, give you a clean summary, no copying or pasting into a chatbot. So, Open Cloud will read the file directly from your disk. Also, organizing a folder. You say, "Sort my download file type." And Open Cloud will do it. Creating a document. You say, "Write meeting's notes from today and save it in notes.md." So, Open Cloud will draft the file based on what you tell and about the meeting, etc. And also like it will save uh the file in your computer. So, the file is on your desk and not in a chat window. Then we have communication. Well, Open Cloud can work with your email, calendar, and your messaging apps. It will It can summarize and read emails. So, you say, "What are five most important unread emails?" And Open Cloud will check your inbox and read the messages, ranks them by importance, and give you a summary. You get the signal without scrolling through everything at once. Then we have drafting the reply. You say, "Draft a reply to Sarah saying I can meet at 2:00 p.m. Tuesday." And Open Cloud will write the reply in a professional tone and then waits for your approval before sending. It never sends anything without your explicit go-ahead. Message June from your phone. So, once Open Cloud is connected to Telegram or WhatsApp, you can text from your phone like a contact, "What meetings do I have tomorrow?" It checks your calendar and replies with the schedule. You do not need to be at your computer. Then we do have like research on web. The This is like the best category because like researching the topic, you can do it by simply saying, "Research the top five projects management tools, compare features and pricing, save it in summary and my desktop." And Open Cloud will search the web, read multiple pages, compare with what it finds, and write a clean summary document. So, when the prop Once the prompt is there is actually like said, you get the results directly on your computer. Automated daily briefing. So, for example, you configure Open Cloud to run every morning at 8:00 a.m. It checks your email, checks your calendar, and sends the Telegram message, "Uh three unread emails, two meetings today, and here are what to do." is actually what matters. So, you wake up to a briefing you did not actually have to ask for. This is actually leading us to understanding a very important point, which is the per- permission system. A natural question at this point would be, "What if Open Cloud does something I did not want to? And what if they list a file or sends an email without asking?" Well, the answer is the permission system. Before Open Cloud will write a file, sends a message, or run any command, it shows you exactly what it wants to do, and waits for your approval. You see the proposed action, you approve or reject, and nothing happens without you saying, uh yes to it. So, you are always in control, and Open Cloud proposes, you decide, and every action is very transparent. Now, those are the kinds of things that Open Cloud can do. Manage files, handle communication, research topics, and automate daily workflows. All from natural language instructions. All with your permissions. Next, we need to understand the core concepts that makes all of this possible. The difference between chatbots and also agent. You have seen what Open Cloud can do. Now, let's understand why it works so differently from the AI tools you already know. A chatbot lives inside a browser tab. You ask it a question, it gives you a text back. That is it. It cannot touch your files, it cannot send emails, it cannot take any real action in the world. An agent runs on your machine. You describe a task, it plans what needs to happen and picks the right tools for you, takes the actions and checks the results and iterate until the job is done. The conversation is not the product and the completed task is the product. Now, there's something very important to understand, the agent tech loop. The agent tech loop itself is basically like what makes OpenCloud follows a three-step loop. Step one, it receives your task and makes a plan deciding which tools it needs. Step two, it acts it acts actually using your tools, reading files, searching the web, calling APIs, and observing what happens. And step three, if the task is not done, it loops back and keeps going until it's done. It reports the results. This loop is what makes an agent an agent. A chatbot does step one and stops. An agent keeps going through all the three steps as many times as needed. Now, something also to understand, persistent memory. When you close ChatGPT, it mostly forgets you. OpenCloud remembers your preference, your projects, things you have told, all sorted as a markdown file on your computer that you can read and edit yourself. And there is also something very important to understand, proactive behavior. Chatbots only respond when you type. Basically, like OpenCloud can reach out to you and set a reminder, schedule a daily check, and ask ask it to basically monitor something. It messages you through Telegram, Slack, or whatever you have connected when there is something to report. That That's what clubs like what we need to understand at the beginning, you know, and you understand what an AI agent is, what OpenCloud does, and how it differs from chatbots. Next, let's move to the installation. Before we install Open Cloud, let's make sure your computer is ready. This video covers the prerequisites. Open Cloud runs on macOS 12 or later, any Mac from 2017 onward, or Windows 10 or 11 64-bit, or Linux Ubuntu 20.04, or any modern distribution. If your computer can run a web browser, it can definitely run Open Cloud. Now, what is Node.js? Node.js is a program that lets your computer run JavaScript code outside a web browser. Open Cloud is built with JavaScript, so it needs Node.js installed to run. You install it once, and then you never have to think about it again. Now, if you open your terminal and type node {dash} {dash} version, in this case I'm on macOS, I will see here the version of the Node.js installed. You can install it on macOS just by going to nodejs.org/en in my case in English. You can install it um like for Windows, and also for Linux as well. The way is very simple. I will click on get Node.js, and we can see here all the platforms where we can install it. Windows, macOS, Linux, and AIX as well. We can uh define which package manager we want to use, or which way we want to install it. We can basically like uh specify the version. So, overall, play with this depending on what you have in your specifications, and just get Node.js. A quick help if you are not like really sure which version you do you have, or what do you want to use exactly to install it, uh by default, you can find here which um like um version is much compatible with yours, with which operating system. In this case, because I have a Mac OS, I'm directly I have it here Mac OS installer.pk pkg. So, I can just like click here and I will have my Node.js installed. As you can see here, it's actually getting installed. At this point, I will click on it and drag it to um like the installer in my applications folder. And eventually, I'm going to follow all the steps and get the um Node.js installed. I already have it, so for your case, you just need to follow the same steps. And the only difference going to be your operating system. Before we install Open Cloud, there is an important decision to make. Where do you want to run it? You have two options and both work well. Let's go through them together. The first option is running Open Cloud locally on your own computer. You install it on your Mac, Windows, or Linux machine. You run it and it stops when you shut down your computer. This is the simplest option and the one most beginners start with. The second option is running Open Cloud on a VPS, which stands for a virtual private server. In this case, Open Cloud runs on a server in the cloud. It stays on 24/7, even when your own computer is actually turned off. You connect to it through a terminal over the internet. Now, the question you might have in your brain, what is a VPS? A VPS is a computer in a data center that you rent by the month. You get your own operating system, your own storage, and full control over everything. The only difference from a regular computer is that you access it remotely through a terminal instead of sitting in front of it. Think of it as renting a computer that never turn off. It runs in a data center somewhere and you connect to it from your laptop whenever you need it. You type commands, it execute them, and everything stays running even after you disconnect. Now, when do we use each? You will use a local installation when you are learning and experimenting. If you are following this course, want to try things out. And don't need your agent running while you sleep. Local is the simplest part. Open cloud runs while your computer is on and stops when you shut down. Now, we will use a VPS when we want to always have an agent. If you want Open Cloud to send you a morning briefing before you wake up, respond to Telegram messages at 3:00 a.m., or monitor something continuously, you need a VPS. The server never turns off, so your agents never stop. Well, overall, we do have few popular VPS providers. We have uh Hetzner, starts at around $4 per month. European data centers, excellent performance for that price, and very popular in the Open Cloud community. We have Digital Ocean, starts at $6 per month, and keen interface, great documentation, beginner-friendly. And also a big cloud provider like AWS and Google Cloud have free tiers for the first year. Though they are like many complex um to set up, and oh it's a bit of overkill for cloud. And eventually, for this case for Open Cloud, Hetzner or Digital Ocean are the recommendations. Uh simple, affordable, and more than enough power for you to start. For the way to go, uh we will pick a provider, Hetzner or like Digital Ocean for beginners. You create an account and add a payment method. You create a server. In this case, you choose Ubuntu 22.0 or as the operating system, and you pick the smallest plan, $8 to $6 a month. And you select the region near to you for the lowest latency. Once the server is created, you receive an IP address and either password or an SSH key. We'll receive this in a bill. Open your terminal on your local machine and you just like type SSH root at your IP address. So, you enter your password when prompted and you are in VPS. From there, you can install Node.js and Open Cloud the same way you would use on a local Linux machine. So overall two things that we need to know. The VPS itself costs $4 to $10 per month. The smallest plan with one CPU and one gigabyte on basically like of RAM is enough for Open Cloud. So, you don't really need a powerful server. On top of the server cost, you still pay for the AI model API usage. So, your total budget for a learning setup is about $10 to $15 per month. Servers plus API calls. For this course, either options works. Me personally, I prefer to always basically work on a VPS for security reasons so I can have everything disconnected from my actual working like desktop. But, if you want the simplest path, you can install it locally first and then you move to the VPS. But, you use the VPS if you want your agent to run around the clock. The installation steps in the next video will work for both. Now that you know where you want to run Open Cloud, let's install it. The next video will cover macOS installation. The one after covers Windows and if you are using a VPS with Ubuntu, follow the macOS videos since the Linux commands are the same.

Assistir ao vídeo original no YouTube

Assuntos: Agentes de IA, Assistente pessoal. Publicado no YouTube em 2026-10-23.

Seleção automática por assunto e idioma. O vídeo pertence ao canal de origem.

About the OpenClaw Gateway, channels and project

OpenClaw is an open-source AI assistant that runs on the user's own computer. According to Wikidata, it is written in TypeScript, released under the MIT license and developed by Peter Steinberger, and it was previously called Clawdbot and Moltbot. The official README says it is stewarded by the OpenClaw Foundation, an independent 501(c)(3) organization, and has no paid tier, hosted service or token. By default, OpenClaw itself only sends a daily version check; anonymous feature statistics are opt-in, and the setting update.checkOnStart: false disables both.

The Gateway is the local control plane for sessions, tools, events and channel connections, and the Control UI, the CLI and the TUI connect to it. The same Gateway can run as a personal assistant on a laptop or as a shared team deployment, with configuration as the only difference. Channels bring the assistant to WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage and other messaging services, and there are native apps for macOS, iOS, Android, Windows and Linux. The official installer supports macOS, Linux and Windows; with npm, the package requires Node 24.16+ or 26.1+ (Node 26 recommended). After onboarding, openclaw gateway status checks the Gateway and openclaw dashboard opens the Control UI. The README advises treating inbound messages as untrusted input; channels that accept direct messages pair unknown senders by default.

Fact sheet

• Name: OpenClaw
• Type: Open-source AI assistant
• Developer: Peter Steinberger
• Former names: Clawdbot and Moltbot
• Language and license: TypeScript, MIT
• Steward: OpenClaw Foundation, an independent 501(c)(3)
• npm requirement: Node 24.16+ or 26.1+ (Node 26 recommended)
• Useful commands: openclaw onboard --install-daemon, openclaw gateway status, openclaw dashboard

Read also

• I quit OpenClaw for this
• Build Your First AI Assistant on a VPS in 15 Minutes With OpenClaw
• How to Use OpenClaw as Your Personal AI Coworker (Beginner Guide)

---

Sources and references: Wikidata – OpenClaw (Q137916205) · GitHub – openclaw/openclaw (README) · video by the Eduonix Learning Solutions channel on YouTube.