How to Install ClawdBot (Openclaw): Setting Up with Telegram & Claude

If you’ve been playing around with Artificial Intelligence lately, you’ve probably noticed a pattern. You open a web browser, navigate to your favorite AI tool, type in a prompt, get an answer, and then close the tab. Once that tab is closed, the conversation is effectively gone—or at least, buried in a history log that you have to log back into a website to see. It feels a bit… transient, doesn’t it? It feels like using a tool rather than working with an assistant.

That is exactly the problem ClawdBot (often referred to in the dev community as Openclaw) is trying to solve.

Imagine an AI that doesn’t just live in a browser tab, but lives directly on your computer. Imagine an assistant that can write code, create files, browse the web, and slide right into your daily workflow via apps like Telegram, Slack, or Discord. That is the promise of ClawdBot.

In this comprehensive, beginner-friendly guide, I’m going to walk you through everything you need to know about ClawdBot. By the end of this article, you will have your own local AI assistant running on your machine, ready to tackle real tasks for you.

So, grab a coffee, and let’s dive into the world of local AI agents.

What Exactly is ClawdBot?

Before we start clicking buttons and typing commands, it’s important to understand what we are installing.

Most AI tools today are “cloud-bound.” You send data to a server, a massive model processes it, and sends text back. ClawdBot flips this script slightly. It is a self-hosted AI assistant. This means the “agent”—the part that talks to you, remembers your context, and executes commands—runs directly on your PC.

However, it still needs a “brain.” ClawdBot connects to powerful AI models (like Claude or ChatGPT) via API keys to get its intelligence. Think of it this way: ClawdBot is the hands and body that lives in your house, and it calls a genius professor (Claude or GPT) on the phone whenever it needs to answer a difficult question.

Why is this better?

1. Persistence: Because the bot runs on your machine, it can interact with your files and your operating system directly.
2. Integration: It doesn’t need a browser open. It can live in your favorite chat apps.
3. Privacy: Your conversations and files sit on your machine, not on a third-party server’s database (though the text is still sent to the AI provider for processing).

It transforms the AI from a “chatbot” into a real assistant that can actually do things.

The Key Features: What Can ClawdBot Actually Do?

When people hear “AI assistant,” they often just think of a chat window. But once you start using ClawdBot, you realize it’s much more. Here is a breakdown of its core capabilities:

• Execute Local Commands: It isn’t just talk; it can run commands on your computer. Need to organize a folder? ClawdBot can do that.
• Write and Create Files: It can generate code, write text documents, create spreadsheets, and even put together presentations directly from your chat.
• Multi-Platform Integration: You aren’t locked into a specific interface. ClawdBot bridges the gap between AI and Telegram, Slack, and Discord. You can talk to it from your phone while your computer does the heavy lifting.
• Web Browsing: If you configure it correctly, it can connect to search engines (like Brave Search) to pull in real-time information, meaning it’s not limited to knowledge from two years ago.
• Memory: Unlike standard chatbots that forget everything when the context window fills up, ClawdBot is designed to remember past interactions and learn your preferences over time.
• Web Dashboard: It comes with a local website (a dashboard) that runs on your computer where you can view chat history, manage settings, and check the bot’s status.

Browser AI vs. ClawdBot: A Quick Comparison

To really drive home the difference, let’s look at a simple comparison table.

Do You Need a Supercomputer (or a Mac)?

One of the most common questions I hear is: “Do I need a Mac Mini or some high-end PC to run this?”

The short answer is no.

Here is the good news: ClawdBot is incredibly lightweight. Because you aren’t running the massive AI model (the “brain”) on your own graphics card, you don’t need a supercomputer. Your computer is just running the “agent” software, which is very light.

You can run ClawdBot on:

• Windows
• macOS
• Linux

There are no specific hardware requirements, and you certainly don’t need to dedicate a whole machine just for this. If your computer can browse the web and run a text editor, it can run ClawdBot.

The Cost Breakdown: Is It Free?

This is where we need to be realistic.

The Software: ClawdBot itself is completely free and open source. You are not paying a subscription fee to the developers of ClawdBot.

The Intelligence: You do need to pay for the API keys. Since ClawdBot connects to models like Claude (Anthropic) or ChatGPT (OpenAI), you need an account with those providers.

• Pay-Per-Use: Most providers work on a “pay-as-you-go” model. You put $5 or $10 into your account, and it charges you pennies for every question you ask.
• Model Choice: The cost depends on which model you choose.
  • Cheaper Models: Faster, less intelligent, great for simple tasks.
  • Premium Models (e.g., GPT-4, Claude Opus): Smarter, better at coding, but burn through credits faster.

Additionally, some advanced features (like specific web search integrations) might require third-party API keys, which could incur small extra costs. But for a beginner just starting out, the cost is negligible—often just a few dollars a month depending on how much you chat.

Step 1: Getting Your “Brain” (The API Key)

Before we install the bot, we need to give it a brain. For this tutorial, I’m going to recommend Claude (Anthropic) because it tends to be excellent at coding and following instructions, but the process is nearly identical if you prefer OpenAI.

1. Sign Up: Go to the Anthropic console (usually console.anthropic.com). If you don’t have an account, create one.
2. Navigate to API Keys: Look for a tab or sidebar menu that says “API Keys” or “Get API Keys.”
3. Create a Key: You’ll see a button to create a new key. Click it.
4. Name It: Give it a name like “ClawdBot” or “My Local Assistant.”
5. Copy It: Crucial Step: Once the key is generated, copy it immediately. You usually won’t be able to see the whole key again for security reasons. Save this in a safe place for a moment.

Note: If you prefer OpenAI, the process is the same—go to platform.openai.com, create a secret key, and copy it.

Step 2: Installing ClawdBot

Now, let’s get the software onto your computer. This is usually the part that scares beginners because it involves the Terminal or Command Prompt. Don’t worry—it’s just copy-pasting text.

Prerequisites

ClawdBot is built on Node.js. If you are a developer, you probably have this installed. If you aren’t, the installer usually handles this for you, but it’s good practice to have Node.js installed beforehand.

The Installation Process

1. Open Your Terminal:
   • On Mac: Press Cmd + Space, type “Terminal,” and hit Enter.
   • On Windows: Press Win + R, type “cmd,” and hit Enter.
2. Get the Command: Head over to the official ClawdBot (Openclaw) website or documentation repository. There will be a specific “one-line install command.” It usually looks something like curl ... or npm install ....
3. Paste and Run: Paste that command into your terminal and hit Enter.

The installer will start working. If you don’t have Node.js or Git installed, the script might detect that and try to help you install them. If you already have them (like many developers do), it will zip right through and just install the ClawdBot agent.

Sit back and let it do its thing. It might take a few minutes to download all the necessary pieces.

Step 3: The Configuration Wizard

Once the installation finishes, the magic happens. ClawdBot doesn’t just install and sit there; it launches directly into a configuration wizard.

You should see some text in your terminal asking if you understand that this software is powerful and inherently risky (because it can control your computer). It will ask you to type “Yes” to continue.

Pro Tip: Always read the warning, but for local use, typing “Yes” is fine to proceed.

Choosing the Onboarding Mode

It will likely ask you to choose an “Onboarding Mode.” For this tutorial, and especially for your first time, choose “Quick Start.”

This option presets a lot of the complicated stuff so you don’t have to manually edit configuration files right out of the gate.

Connecting the Brain

Next, the wizard will ask you to choose your AI Provider.

• Select Anthropic (since we grabbed that key earlier).
• It will ask for your API Key. Paste that long string you copied in Step 1.

Selecting a Model

Now you have to pick the specific model. This is a balance of cost vs. intelligence.

• Claude 3 Haiku: Fast, cheap, good for simple tasks.
• Claude 3 Sonnet: The middle ground. Great balance of speed and smarts. (I recommend this for beginners).
• Claude 3 Opus: The genius. Slower and more expensive, but incredible for complex coding.

Select Sonnet (or the current equivalent medium-tier model) to start. It gives you the best bang for your buck. Remember, you aren’t locked into this choice forever; you can change it in a config file later.

Step 4: Bringing It to Telegram

This is where it gets cool. You don’t want to stare at a terminal window to chat. You want to chat on your phone while you’re making coffee. Let’s connect it to Telegram.

Setting Up the Telegram Bot

1. Open Telegram and search for BotFather. It’s the official bot for making other bots.
2. Start a chat with BotFather and send the command /newbot.
3. BotFather will ask you to name your bot. Pick something fun, like “MyLocalBot.”
4. Next, it will ask for a username. This must end in bot, like MyLocalBot_bot or JohnsAI_bot.
5. Once you’ve done that, BotFather will give you a HTTP API Token. It looks like a long string of random characters. Copy this.

Linking ClawdBot to Telegram

Go back to your terminal where ClawdBot is running. The configuration wizard (or the settings menu) will ask if you want to connect Telegram.

• Select “Yes” or “Configure Telegram.”
• Paste the token you got from BotFather.

The terminal should show a success message indicating that ClawdBot is now listening for messages from that specific Telegram bot.

Step 5: Your First Conversation

Everything is set up. Now for the moment of truth.

1. Open Telegram on your phone or desktop.
2. Search for the username of the bot you just created (e.g., MyLocalBot_bot).
3. Open the chat and type /start.

After a brief second, you should get a reply. It might say something like, “ClawdBot is connected and ready.”

You are now talking to your own local AI assistant, hosted on your computer, via your phone.

Try saying hello. Ask it a general knowledge question. You’ll notice it feels just like chatting with ChatGPT, but remember—this is running through your local infrastructure.

Building Context

The first thing you should do is tell it who you are. The more context you give it, the better it assists you.

• Example Prompt: “Hi, I’m a beginner web developer learning Python. I like concise answers. Please remember this for future sessions.”

ClawdBot will save this preference. As you use it more, it builds a memory of your projects and your style.

Exploring the Features: What to Do Next?

Now that you have it running, here is how you can actually use it for work.

1. Coding Assistance

Since ClawdBot lives on your machine, you can actually have it write code for you.

Example: You are working on a website and need a button.

• You: “Write the HTML and CSS for a blue, rounded button that turns red when I hover over it.”

ClawdBot will generate the code. You can copy-paste it, or if you get more advanced, you can configure it to write directly to files on your desktop.

2. File Management (The “Agent” Part)

This is where it separates itself from browser AI. You can ask it to interact with your file system.

Example:

• You: “Create a new folder on my desktop called ‘Project Alpha’ and inside it, create a text file named ‘notes.txt’ containing a summary of our last chat.”

Note: For this to work, you may need to grant the terminal or the specific user permissions to write to those folders, but ClawdBot handles the execution of the commands.

3. Web Research

If you configured a search provider (like Brave Search) during the setup, ClawdBot can look things up for you.

Example:

• You: “Search the web for the latest Python version released this year and summarize the changes for me.”

It will run a search, read the results, and give you a human-friendly summary.

4. The Web Dashboard

While Telegram is great for chatting, sometimes you want a bird’s-eye view. ClawdBot runs a local dashboard on your computer.

1. Open your web browser.
2. Type in the address: http://localhost:18789 (or whatever port the setup specified).
3. Here, you can see your full chat history, the status of the bot, and configuration settings.

This is incredibly useful if you sent a long command from your phone but want to read the detailed response on a large screen.

Understanding the “Memory” Feature

One of the standout features discussed earlier is memory. Browser AI is often “stateless”—if you close the tab, it forgets what you were working on (unless it has a limited history sidebar).

ClawdBot is designed to be persistent. Because the database is on your machine, it can reference conversations from days or weeks ago.

How to leverage this:
Whenever you start a new project, have a “setup session” with the bot.

• You: “I’m starting a new project about a coffee shop blog. Here are the requirements…”

By establishing this context early, you can simply ask days later, “Rewrite the ‘About Us’ page we discussed,” and it will know exactly what you are talking about.

Coding Examples (A Sneak Peek)

For those interested in the technical side, here is a simple example of how you might interact with ClawdBot for coding.

Scenario: You need a simple Python script to organize files by extension.

Prompt to ClawdBot:

“Write a Python script that looks at the files in my current directory, creates folders for ‘Images’, ‘Documents’, and ‘Others’, and moves the files into the correct folders based on their extensions.”

ClawdBot’s Response might look like this:

import os
import shutil

# Define the categories and their extensions
categories = {
    "Images": [".jpg", ".jpeg", ".png", ".gif"],
    "Documents": [".pdf", ".docx", ".txt", ".xlsx"],
}

# Create directories
for folder in categories:
    if not os.path.exists(folder):
        os.makedirs(folder)

# Move files
for filename in os.listdir("."):
    if os.path.isfile(filename):
        ext = os.path.splitext(filename)[1].lower()
        moved = False
        for folder, extensions in categories.items():
            if ext in extensions:
                shutil.move(filename, os.path.join(folder, filename))
                moved = True
                break
        if not moved:
            if not os.path.exists("Others"):
                os.makedirs("Others")
            shutil.move(filename, os.path.join("Others", filename))

print("Files organized successfully!")

import os
import shutil

# Define the categories and their extensions
categories = {
    "Images": [".jpg", ".jpeg", ".png", ".gif"],
    "Documents": [".pdf", ".docx", ".txt", ".xlsx"],
}

# Create directories
for folder in categories:
    if not os.path.exists(folder):
        os.makedirs(folder)

# Move files
for filename in os.listdir("."):
    if os.path.isfile(filename):
        ext = os.path.splitext(filename)[1].lower()
        moved = False
        for folder, extensions in categories.items():
            if ext in extensions:
                shutil.move(filename, os.path.join(folder, filename))
                moved = True
                break
        if not moved:
            if not os.path.exists("Others"):
                os.makedirs("Others")
            shutil.move(filename, os.path.join("Others", filename))

print("Files organized successfully!")

What makes this special?
With a browser bot, you would copy this code, open your text editor (like VS Code), create a file called organize.py, paste it, save it, and then run it in your terminal.

With ClawdBot, because it is an agent, you can (in advanced setups) ask it to “Save that script to my desktop as organize.py and run it.” It cuts out the middleman (you) for the repetitive tasks.

Troubleshooting Common Issues

Let’s be real—setting up local software doesn’t always go perfectly the first time. Here are some common bumps in the road and how to smooth them out.

“I can’t connect to Telegram.”

• Check: Did you copy the correct token from BotFather? It’s easy to mix up tokens if you have multiple bots.
• Check: Is your terminal window still open? ClawdBot needs to be running in the terminal to listen for messages. If you close the terminal, the bot goes to sleep.

“It’s giving me an API Error.”

• Check: Did you paste the API key correctly?
• Check: Do you have credits in your Anthropic or OpenAI account? Sometimes accounts run out of funds if you’re on a tight budget.
• Check: Is your internet connection working? Remember, the brain is in the cloud, even if the body is local.

“The terminal commands look scary.”

• Solution: Don’t panic. If the installer stops and asks you a question (Y/N), just read it. Usually, typing “Y” (Yes) is the right move to proceed with the installation.

Advanced Tips for Power Users

Once you’ve gotten comfortable with the basics—chatting on Telegram and generating code—here are a few ways to level up:

1. Switch Models on the Fly

Don’t feel locked into the model you chose during setup. If you have a complex coding task, you might temporarily switch to a more powerful model (like GPT-4 or Claude Opus) for that specific session, then switch back to a cheaper one for casual chatting. You can usually do this via a command in the chat, like /model gpt-4.

2. Privacy Mode

If you are working with sensitive data (like personal finance info), remember that while the files are on your PC, the prompts are sent to the AI. Most major providers have policies that say they don’t train on data sent via API, but if you are paranoid, you might want to use a local LLM (like running Llama 3 on your own GPU) instead of an API key. ClawdBot supports various setups, though that is a more advanced configuration.

3. Integration with Daily Workflow

Set up specific “Channels” or “Groups” in Slack or Discord dedicated to ClawdBot.

• Example: Create a Slack channel called #ai-dev-helper. Add ClawdBot to that channel. Now, your whole team can ask the local bot questions, and it can access the shared codebase if you give it permission.

WrapUP: Is It Worth It?

Setting up ClawdBot is definitely more work than just opening a tab in Chrome. You have to deal with terminals, API keys, and BotFather. It takes about 30 to 60 minutes to get everything running smoothly.

But the payoff is huge.

You move from being a “user” of AI to an “owner” of an AI assistant. You have a tool that:

• Respects your privacy (mostly).
• Integrates into your apps.
• Remembers your work.
• Can actually do work on your computer.

It transforms AI from a novelty into a genuine productivity engine. Whether you are a developer, a writer, or just someone who loves tinkering with tech, ClawdBot offers a glimpse into the future of personal computing—a future where your computer doesn’t just wait for your input, but actively helps you achieve your goals.

So, if you’re tired of the browser-tab cycle and ready for an assistant that feels like it’s actually yours, give ClawdBot a try. Once you experience the freedom of chatting with your own computer from your phone, you won’t want to go back to the old way.

References

For further reading and official downloads, you can check out these resources:

1. Node.js Official Website
2. Telegram BotFather Guide

FAQs