In the rapidly evolving landscape of AI-powered assistants, having your agents accessible where you work and communicate is paramount. For many developers and technical users, Telegram stands out as the channel of choice for OpenClaw deployments in 2026. Its robust API, ubiquitous availability across devices, and inherent speed make it an ideal platform for interacting with your AI agent, whether you’re debugging, querying data, or automating tasks.

This guide will walk you through setting up your OpenClaw agent on Telegram from scratch, focusing on practical steps, configuration examples, and real-world use cases relevant to a developer’s workflow. We’ll move beyond the basics to ensure your setup is not just functional, but also secure and ready for your day-to-day operations.

Prerequisites

Before diving in, ensure you have the following:

• An active Telegram account.
• Access to an OpenClaw instance (either a cloud-hosted service or a self-managed deployment). For this guide, we’ll assume you have the `openclaw` CLI tool installed and configured to connect to your instance.
• Basic familiarity with the command line interface (CLI).

Step 1: Creating Your Telegram Bot with BotFather

The first step involves creating a new bot on Telegram itself. This is handled by the official BotFather bot, a Telegram service dedicated to managing bots. It’s straightforward and provides you with the crucial API token needed to connect OpenClaw.

Open your Telegram app and search for @BotFather. Start a chat with it. Once you’re in the chat, send the command:

/newbot

BotFather will then prompt you for two pieces of information:

1. Choose a name for your bot: This is the display name that users will see in their chat list. It can be descriptive and user-friendly, e.g., “OpenClaw Dev Assistant” or “Project Phoenix Bot”. Let’s use “OpenClaw Tech Buddy” for our example.
2. Choose a username for your bot: This must be unique across all of Telegram and must end with “bot” (case-insensitive). This is how users will find your bot (e.g., @myopenclaw_buddy_bot). A good practice is to make it memorable and relevant to its function. For our example, let’s go with OpenClawTechBuddy_bot. If the username is taken, BotFather will prompt you to choose another one.

Upon successful creation, BotFather will send you a message containing your bot’s API token. This token is a long string of alphanumeric characters (e.g., 1234567890:AABBCCDDeeFFggHHiiJJkkLLmmNNOOPP). This token is sensitive; treat it like a password. Do not share it publicly or commit it directly to version control. Copy this token and save it securely for the next step.

BotFather will also provide a link to your new bot (e.g., t.me/OpenClawTechBuddy_bot), which you can use later to start a conversation.

Step 2: Configuring OpenClaw for Telegram Integration

With your Telegram bot token in hand, it’s time to tell your OpenClaw instance how to connect. OpenClaw offers flexible configuration options, whether through an interactive setup, environment variables, or a dedicated configuration file. We’ll cover the most common methods.

Using the Interactive Setup (Recommended for First-Timers)

If you’re running OpenClaw for the first time or want to quickly update its channel, the interactive setup is the easiest path:

openclaw setup

The CLI will guide you through various configuration steps. When prompted for the communication channel, select “Telegram”. You’ll then be asked to provide your bot token:

Which channel would you like to configure?
1. Web Chat
2. Telegram
3. Slack
> 2

Please enter your Telegram Bot API Token:
> [PASTE_YOUR_TELEGRAM_BOT_TOKEN_HERE]

OpenClaw will now attempt to connect to Telegram. This might take a moment...
Connection successful! Your OpenClaw agent is now configured for Telegram.

OpenClaw handles the underlying complexities of setting up webhooks or long-polling with Telegram’s API, ensuring a smooth connection.

Advanced Configuration (Environment Variables or Config File)

For production deployments, Docker containers, or CI/CD pipelines, using environment variables or a configuration file is often preferred for automation and consistency.

Environment Variables

You can set the Telegram token as an environment variable before starting your OpenClaw instance:

export OPENCLAW_CHANNEL="telegram"
export OPENCLAW_TELEGRAM_BOT_TOKEN="YOUR_TELEGRAM_BOT_TOKEN_HERE"

# Then start your OpenClaw instance
openclaw start

This approach is excellent for Docker setups, where you can pass these variables directly:

docker run -d \
  -e OPENCLAW_CHANNEL="telegram" \
  -e OPENCLAW_TELEGRAM_BOT_TOKEN="YOUR_TELEGRAM_BOT_TOKEN_HERE" \
  openclaw/openclaw-agent:latest

Configuration File (e.g., openclaw.yaml)

For more complex setups, you might manage your OpenClaw configuration in a YAML file. Create or edit an openclaw.yaml file in your OpenClaw working directory:

# openclaw.yaml
channel: "telegram"
telegram:
  bot_token: "YOUR_TELEGRAM_BOT_TOKEN_HERE"
  # Optional: Configure webhook options if desired for advanced scenarios
  # webhook_url: "https://your-public-openclaw-endpoint.com/telegram"
  # webhook_port: 8443
  # allowed_updates: ["message", "edited_message", "callback_query"]

Then, start OpenClaw referencing this configuration file:

openclaw start --config openclaw.yaml

Remember to replace YOUR_TELEGRAM_BOT_TOKEN_HERE with the actual token you obtained from BotFather.

Step 3: Initializing and Testing Your OpenClaw Bot

Once OpenClaw is configured and running, it’s time to bring your agent to life on Telegram. Find your bot using the username you chose earlier (e.g., @OpenClawTechBuddy_bot) in Telegram’s search bar.

Open a chat with your bot and send the /start command:

/start

This initiates the conversation and usually prompts a welcome message from your OpenClaw agent. If you receive a message like “Hello! I’m your OpenClaw agent, ready to assist you. How can I help?”, congratulations! Your setup is complete, and your OpenClaw agent is now live on Telegram.

Troubleshooting Common Issues

• Bot not responding:
  • Double-check that your OpenClaw instance is running.

  Written by: Alex Torres, Editor at OpenClaw Resource

  Last Updated: May 2026

  Our Editorial Standards | How We Review Skills | Affiliate Disclosure

  Want to see what OpenClaw can really do? Check out this wild project building AI agents with physical bodies →

How to Use OpenClaw with Ollama for Local AI (No Cloud Required)

As developers, we’re constantly pushing the boundaries of what’s possible with AI. But often, that comes with trade-offs: API costs, data privacy concerns, and reliance on external services. What if you could harness the power of large language models (LLMs) for your AI agents without any of those compromises? Enter OpenClaw and Ollama – a powerful combination that lets you run sophisticated AI agents entirely on your local hardware, keeping your data, costs, and control firmly in your hands.

This guide will walk you through setting up OpenClaw to leverage Ollama as its local AI backend. We’ll cover everything from hardware considerations to practical configuration, ensuring you can build intelligent agents that operate with unparalleled privacy and efficiency.

Understanding the Pillars: Ollama and OpenClaw

What is Ollama? Your Local LLM Server

Think of Ollama as your personal, lightweight server for large language models. It takes the inherent complexity of running models like Llama 3, Mistral, or Gemma – handling everything from model quantization and loading to managing GPU acceleration – and boils it down to a simple command-line interface and an accessible API endpoint. Instead of needing to wrangle with deep learning frameworks, you simply tell Ollama which model you want, and it makes it available locally.

For OpenClaw, Ollama becomes the direct replacement for cloud-based LLM providers like OpenAI’s GPT or Anthropic’s Claude. It serves as the engine that powers your agent’s reasoning, understanding, and generation capabilities, all from your machine.

What is OpenClaw? Your Agentic Framework

OpenClaw is an open-source framework designed for building robust and intelligent AI agents. It provides the structure for defining agent roles, tools, memory, and execution flows. While OpenClaw is designed to be model-agnostic, supporting various cloud LLM providers out of the box, its true power for many developers lies in its flexibility to integrate with local models. By connecting OpenClaw to Ollama, you empower your agents with the ability to perform complex tasks, analyze data, and generate content without sending a single byte of sensitive information beyond your local network.

The “No Cloud” Advantage: Why Go Local?

Running your OpenClaw agents with Ollama isn’t just a technical exercise; it’s a strategic choice that offers significant advantages:

• Unmatched Data Privacy & Security: This is arguably the biggest benefit. Your sensitive code, proprietary data, or confidential client information never leaves your machine. This is crucial for industries like healthcare, finance, or defense, and for any developer working with private datasets.
• Zero API Costs: Say goodbye to fluctuating monthly bills for token usage. Once your hardware is acquired, the operational cost of running models locally is effectively zero, making long-running or high-volume agent tasks far more economical.
• Offline Capability: Develop and deploy agents in environments without internet access – ideal for fieldwork, secure intranets, or simply working from a remote cabin.
• Complete Control & Customization: You’re not beholden to a third-party API’s rate limits, model updates, or downtime. You choose which models to run, when to update them, and can even fine-tune models directly on your hardware for highly specialized tasks.
• Reduced Latency: For many tasks, especially those involving rapid iteration or real-time interaction, keeping the LLM inference loop local can significantly reduce latency compared to round-trips to cloud APIs.

Hardware Requirements: The Practicalities of Local AI

While the “no cloud” promise is appealing, local AI does have hardware prerequisites, primarily centered around RAM and GPU capabilities. The good news is that modern hardware, especially Apple Silicon Macs and NVIDIA GPUs, are increasingly capable.

• RAM is Key: LLMs consume RAM proportional to their size (number of parameters). Generally, you need RAM roughly equal to the model size plus some overhead for the operating system and other applications.
  • Llama 3.1 8B: ~8-10GB RAM (Excellent quality/speed balance for most dev tasks. A modern MacBook Pro with 16GB unified memory handles this well.)
  • Mistral 7B: ~8-10GB RAM (Fast, efficient, and often outperforms larger models in specific benchmarks. Great starting point.)
  • Llama 3.1 70B: ~40-50GB RAM (For cutting-edge quality and complex reasoning. Requires high-end hardware like a Mac Studio M2 Ultra (64GB+) or a desktop PC with an NVIDIA RTX 4090/4080 Super with 24GB VRAM.)
  • Phi-3 Mini 3.8B: ~4-6GB RAM (Extremely fast, good for simpler tasks or constrained environments. Runs well on a Mac Mini M2 with 8GB RAM.)
• GPU Acceleration (Highly Recommended): While Ollama can run models on CPU, a dedicated GPU or integrated neural engine (like Apple Neural Engine) dramatically speeds up inference.
  • Apple Silicon: M1, M2, M3, M4 chips (Pro, Max, Ultra variants) are exceptional due to their unified memory architecture and powerful neural engines. A MacBook Pro M3 Pro with 18GB unified memory is a fantastic sweet spot for 7B-13B models.
  • NVIDIA GPUs: For Windows and Linux desktops, NVIDIA’s RTX series (30-series, 40-series) are the gold standard. More VRAM is always better. An RTX 4060 (8GB VRAM) can handle smaller models, while an RTX 4080 Super (16GB VRAM) or RTX 4090 (24GB VRAM) opens up possibilities for larger models.

Practical Tip: Start with a smaller model like Mistral 7B or Llama 3.1 8B. They offer a great balance of performance and quality without demanding top-tier hardware.

Step-by-Step Setup: OpenClaw with Ollama

Step 1: Install Ollama

First, get Ollama up and running on your system.

macOS & Linux:

Open your terminal and run:

curl -fsSL https://ollama.com/install.sh | sh

This script will download and install Ollama. Once installed, it will automatically start a background service.

Windows:

Download the installer directly from the Ollama website and follow the on-screen instructions. Ollama will install as a service and start automatically.

You can verify Ollama is running by opening a new terminal and typing:

ollama

It should display a list of available commands.

Step 2: Pull an LLM with Ollama

Now, let’s download a model. For this example, we’ll use Llama 3.1 8B, a powerful and versatile model. Feel free to substitute with `mistral`, `gemma:2b`, or `codellama` if you prefer.

ollama pull llama3.1:8b

This command will download the model. It might take a while depending on your internet connection, as these models can be several gigabytes in size. Once downloaded, the model is cached locally and ready for use.

You can test the model directly from the terminal:

ollama run llama3.1:8b

Type a prompt like “Explain quantum entanglement in simple terms.” and press Enter. You should get a response from your local LLM.

Step 3: Install OpenClaw

It’s always a good practice to use a virtual environment for Python projects to manage dependencies cleanly.

Want to see what OpenClaw can really do? Check out this wild project building AI agents with physical bodies →

How Much RAM Does OpenClaw Need? (2026 Guide)

One of the most frequent questions we get from new OpenClaw users, and even seasoned veterans scaling up their operations, is about RAM. How much do you really need? The answer, as always in the technical world, is “it depends.” But let’s be blunt: that’s not helpful enough. This guide aims to give you concrete, practical advice, complete with real-world scenarios, command examples, and a clear breakdown for every major use case in 2026.

OpenClaw is designed to be lean and modular. Its core function is to orchestrate AI tasks, manage conversational states, route requests, and integrate various tools and models. This means its own footprint is surprisingly small, but the resources it *orchestrates* can be incredibly hungry. Understanding this distinction is key to sizing your system correctly.

OpenClaw’s Core Footprint: The Foundation

Let’s start with the OpenClaw process itself. As a Node.js application, OpenClaw is remarkably efficient at idle. You’ll find the core process typically consuming around 200-400MB of RAM. This covers the runtime, its loaded modules, and basic operational overhead. This is your baseline, regardless of what you connect it to.

For a bare-bones installation, where OpenClaw is the only significant application running on a lightweight Linux distribution, you could technically get away with 512MB of RAM. However, this leaves virtually no room for the operating system, file caching, or any other background processes. We strongly advise against this for anything beyond initial testing.

A more realistic minimum for a stable, responsive OpenClaw instance is 1GB of RAM. This provides comfortable headroom for the OS (e.g., a modern Linux server install might take 300-500MB), the OpenClaw process, and a bit of buffer for minor system tasks. If you’re running OpenClaw inside a Docker container, this 1GB minimum still applies to the container’s allocated memory.

# Example: Checking OpenClaw's RAM usage on Linux
# First, find the PID of your OpenClaw Node.js process
ps aux | grep openclaw | grep -v grep

# Let's say the PID is 12345
# Use 'htop' for an interactive view or 'smem' for more detail (install if needed)
htop -p 12345
# Or, for a quick check:
cat /proc/12345/status | grep VmRSS

Scenario 1: Cloud-Based AI Models Only (Lean & Agile)

This is the simplest and often most cost-effective scenario. You’re using OpenClaw as a sophisticated routing layer and state manager, but all the heavy AI lifting is done by external APIs like OpenAI’s GPT-4, Anthropic’s Claude 3, or Google’s Gemini. Your system just sends requests and receives responses.

For this use case, 2GB of RAM is more than sufficient. This covers:

• OpenClaw core process (~300MB)
• Operating system overhead (~500MB for a server OS like Ubuntu Server or AlmaLinux)
• Headroom for network operations, SSL handshakes, and buffering API responses
• Potential light browser automation (e.g., a headless Chrome instance for a quick login before making an API call), which might temporarily spike usage.

Real-world example: You’re building an internal customer support bot that routes queries to Claude 3 Opus, stores conversation history in a PostgreSQL database (hosted externally), and uses OpenClaw to handle user interaction and tool calling. A DigitalOcean Droplet with 2GB RAM (e.g., Basic Droplet, $14/month for 2vCPUs, 2GB RAM) or an AWS t2.small (2GB RAM, ~$15-20/month) would easily handle this for a moderate load of concurrent users.

Scenario 2: Standard Use with Local Operations (The Sweet Spot)

Most OpenClaw users fall into this category. You’re leveraging cloud models, but also performing local tasks like file processing, data transformation, simple database interactions, or more involved browser automation. This is where OpenClaw starts to shine as a true automation platform.

For these mixed workloads, 4GB of RAM is the sweet spot. This provides ample room for:

• OpenClaw process and OS
• Multiple concurrent tasks (e.g., parsing several large JSON files, uploading results to S3)
• Robust browser automation (e.g., using Puppeteer or Playwright to scrape data from a complex website, which can be RAM-intensive, often consuming 100-300MB per browser instance)
• Running lightweight local services or micro-databases (e.g., SQLite, Redis for caching)

Real-world example: An OpenClaw agent that takes user input, uses GPT-4 for initial classification, then scrapes product data from an e-commerce site using Playwright, processes the scraped HTML locally to extract specific fields, and finally stores the structured data in a local SQLite database before reporting back. Here, the Playwright browser instance and data processing will be the primary RAM consumers beyond OpenClaw itself. A VM with 4GB RAM (e.g., an AWS t3.medium or a slightly larger DigitalOcean Droplet) offers excellent value for this kind of work.

Scenario 3: Integrating Local AI Models (The RAM Hogs)

This is where RAM requirements can skyrocket, and it’s also where OpenClaw offers tremendous power and cost savings for specific tasks. Running Large Language Models (LLMs) locally, especially for CPU inference, is extremely RAM-intensive because the entire model weights, plus the context window, must be loaded into memory.

The key factor here is the model size (e.g., 7B, 13B, 70B parameters) and its quantization level (e.g., Q4_K_M, Q8_0). Quantization reduces the precision of the model weights, making them smaller and faster, but with a slight hit to quality. OpenClaw typically integrates with local models via inference engines like Ollama or Llama.cpp.

Sub-Scenario 3a: Small Local Models (7B-13B Quantized)

For models like LLaMA-3 8B Q4_K_M or Mi

Want to see what OpenClaw can really do? Check out this wild project building AI agents with physical bodies →

OpenClaw for Small Business Owners: A Practical Guide

Running a small business means wearing every hat — sales, operations, customer service, marketing, bookkeeping. There’s never enough time. OpenClaw is an AI agent platform that can take a meaningful chunk of the admin workload off your plate, letting you focus on the work that actually moves the needle.

This guide is written for business owners who are not technical. No jargon. No coding. Just practical use cases and how to get started.

What OpenClaw Can Do for Your Business

Think of OpenClaw as a virtual assistant that never sleeps, learns your preferences over time, and can handle a wide range of tasks via simple text messages. Here’s what small business owners are actually using it for:

Customer Communication

• Draft responses to customer inquiries (you review and send, or it sends automatically)
• Follow up with leads who haven’t responded after a set number of days
• Send appointment reminders via Telegram or email
• Respond to common questions with pre-approved answers

Daily Briefings

• Get a morning summary of unread emails and priority items
• See a daily rundown of your calendar and upcoming deadlines
• Receive alerts for urgent client messages or time-sensitive issues

Content and Marketing

• Draft social media posts for the week in one sitting
• Repurpose content (a blog post → email newsletter → three social posts)
• Research competitors and summarize what they’re doing
• Write first drafts of blog posts or case studies

Research and Reporting

• Research vendors, suppliers, or potential partners before a call
• Summarize industry news relevant to your business
• Track specific topics or competitor mentions online
• Compile weekly or monthly performance summaries

Internal Operations

• Create and manage to-do lists and project notes
• Draft SOPs (standard operating procedures) from your voice notes
• Organize files and documents in your workspace folder
• Set reminders and automated check-ins for recurring tasks

Real-World Scenarios by Business Type

Freelance Consultant or Coach

Before client calls, ask your agent to research the client’s business, recent news, and any notes from previous conversations. Ask it to draft a follow-up email after the call. Have it remind you to invoice clients at month-end. This alone can save 3–5 hours per week.

Retail Store (Online or Physical)

Monitor supplier websites for inventory changes. Draft product descriptions. Respond to common customer questions. Track your best-selling products by reviewing sales data files. Get a daily summary of order volume and issues.

Restaurant or Food Business

Draft responses to reviews (you approve before posting). Create weekly specials posts for social media. Monitor reservation requests from email. Draft staff schedule reminders. Track local event calendars for catering opportunities.

Real Estate Agent

Get alerts when new listings match client criteria (if you configure web monitoring). Draft personalized follow-up emails for leads. Summarize recent property market trends from news feeds. Create social media posts for new listings.

Trades Business (Plumber, Electrician, Contractor)

Draft quotes and follow-up messages. Organize job notes and client history. Send appointment confirmation reminders via Telegram. Track material costs by summarizing supplier invoices you send as images.

How to Get Started (Non-Technical Version)

You don’t need to be technical to get OpenClaw running. Here’s the simple path:

Option A: DIY Setup

1. Follow our 30-minute setup guide
2. Host it on a cheap cloud server (about $4–6/month on DigitalOcean or Vultr)
3. Connect it to Telegram for mobile access
4. Spend 30 minutes customizing the USER.md and SOUL.md files to tell it about your business

Option B: Hire Someone to Set It Up

If you’d rather not touch any of the technical setup, there are freelancers on Upwork and Fiverr who specialize in OpenClaw configuration. A basic setup service typically runs $150–$400 and includes configuration, Telegram connection, and a brief training session.

What to Tell Your Agent About Your Business

The more context your agent has, the more useful it becomes. Edit the USER.md file in your workspace to include:

• Your name and your business name
• What you do and who your customers are
• Your working hours and timezone
• Your communication style (formal vs. casual)
• Your most common tasks and pain points
• Key contacts (important clients, suppliers, team members)
• Tools you use (CRM name, email platform, project management tool)

Think of this like onboarding a new employee. The more you explain upfront, the faster they become effective.

Costs: What to Expect

Running OpenClaw for a small business typically costs:

• OpenClaw software: Free (open-source)
• AI model (Claude API): $5–$20/month depending on usage
• VPS hosting (for 24/7 operation): $4–$6/month
• Total: ~$10–$26/month

Compare that to a part-time virtual assistant ($15–$25/hour) or a no-code automation tool like Zapier ($20–$49/month) — and the value is obvious.

Limitations to Know About

OpenClaw is powerful, but it’s not magic. Be realistic about what it can and can’t do:

• It makes mistakes. Always review important customer-facing content before it goes out
• It needs good instructions. Vague requests get vague results. Be specific
• It’s not a CRM. It’s not going to replace Salesforce or HubSpot for complex sales tracking
• Setup takes time. The first few weeks involve a lot of tweaking as your agent learns your preferences

The Business Owner’s Quick-Start Checklist

• ☐ Install OpenClaw and connect Telegram
• ☐ Edit USER.md with your business context
• ☐ Set up HEARTBEAT.md with daily check-in tasks
• ☐ Test with 5–10 real tasks from your workday
• ☐ Identify the 3 most time-consuming repeatable tasks and configure the agent to handle them
• ☐ After 2 weeks: evaluate what’s working and refine

Resources

• OpenClaw Setup Guide
• Telegram Setup Guide
• OpenClaw Skills Guide
• Best Hosting for OpenClaw
• How to Make Money with OpenClaw

The small business owners getting the most value from AI agents right now are the ones who jumped in early, accepted a few weeks of learning curve, and built workflows that match their actual business. Start small, stay consistent, and let the agent prove its value over time.

Recommended on Amazon: Homelab Book | Linux Command Line Book

Want to see what OpenClaw can really do? Check out this wild project building AI agents with physical bodies →

OpenClaw Telegram Setup: Complete Guide

Telegram is the primary way most people interact with their OpenClaw agent. It turns your AI assistant into a mobile-friendly chat interface — you can send it tasks, receive proactive updates, and control your agent from anywhere using just your phone. This guide covers everything from creating your bot to advanced configuration.

Why Telegram?

OpenClaw supports multiple communication channels, but Telegram is the most popular for good reasons:

• Free: Telegram is completely free with no ads
• Fast: Messages are delivered almost instantly
• Cross-platform: Works on iPhone, Android, Windows, Mac, and browser
• Bot API: Telegram has an excellent, well-documented bot system
• Secure: End-to-end encryption available
• No phone number required for bots: Your bot communicates privately without exposing personal info

Prerequisites

Before starting, make sure you have:

• OpenClaw installed and configured (see our Setup Guide if not)
• A Telegram account (download at telegram.org or your app store)
• Your Anthropic API key already configured in OpenClaw

Step 1: Create Your Telegram Bot

Every OpenClaw agent needs its own Telegram bot. Creating one is free and takes about 2 minutes:

1. Open Telegram and search for @BotFather (the official bot creation service from Telegram)
2. Start a conversation and send: /newbot
3. BotFather will ask for a name — this is the display name (e.g., “My Assistant”)
4. Then it will ask for a username — must end in “bot” (e.g., “myassistant_bot” or “john_agent_bot”)
5. BotFather will send you a bot token — a long string that looks like 1234567890:ABCdefGHIjklMNOpqrSTUvwxyz
6. Copy and save this token somewhere safe — treat it like a password

Important: Keep your bot token private. Anyone with the token can control your bot.

Step 2: Install the Telegram Plugin

In your terminal, install the OpenClaw Telegram plugin:

openclaw plugin install telegram

The installer will prompt you for your bot token. Paste it in and press Enter.

Step 3: Configure Your Chat ID

For security, you’ll want to restrict your bot to only respond to messages from you (or your team). This requires your Telegram user ID.

To find your user ID:

1. Search for @userinfobot on Telegram
2. Send it any message
3. It will reply with your numeric user ID (e.g., 123456789)

Add this to your OpenClaw Telegram plugin configuration so the bot ignores messages from anyone else.

Step 4: Start OpenClaw and Test

Start your OpenClaw agent:

openclaw start

Now go to Telegram, find your bot (search by username), and start a conversation. Send a simple message like:

Hello, are you there?

Your agent should respond within a few seconds. If it does — you’re set up correctly!

Understanding Your Bot’s Behavior

Once connected, you can interact with your OpenClaw agent just like you would with any Telegram chat:

• Send messages asking it to do things: “Search for the weather in New York”
• Give it tasks: “Create a new file called meeting-notes.md with today’s date”
• Ask questions: “What did we talk about last Tuesday?”
• Use slash commands: /status, /reset, /reasoning

Setting Up Proactive Notifications

One of OpenClaw’s best features is that it can message you — without you asking first. This is called a heartbeat or proactive notification.

To configure this, edit your HEARTBEAT.md file in your workspace:

• List things you want the agent to check periodically (emails, calendar, weather)
• Set quiet hours so it doesn’t disturb you at night
• Define when it should reach out vs. stay silent

Example heartbeat checklist:

- Check email for urgent messages
- Check calendar for events in the next 2 hours
- Check weather if it's morning (6-10am)
- Stay quiet between 11pm and 8am

Using OpenClaw in Telegram Groups

You can add your OpenClaw bot to Telegram group chats — useful if you want to share your agent with a small team or family.

To add your bot to a group:

1. Open the group in Telegram
2. Tap the group name → Edit → Add Members
3. Search for your bot’s username and add it
4. Make the bot an admin if you want it to read all messages (required for some features)

Privacy tip: In groups, OpenClaw will see all messages. Configure it to only respond when directly mentioned (@yourbotname) to avoid it replying to every conversation.

Telegram Commands Reference

These commands work when sent to your OpenClaw bot in Telegram:

• /start — Initiate conversation with the bot
• /help — List available commands
• /status — See agent status and active settings
• /reset — Clear conversation context
• /reasoning — Toggle extended thinking mode
• /stop — Stop a running task
• /approve allow-once — Approve a pending action
• /approve deny — Deny a pending action

Troubleshooting Common Issues

Bot Doesn’t Respond

• Check that OpenClaw is still running (openclaw status)
• Verify the bot token is correct in your configuration
• Make sure you’re messaging the right bot (search by exact username)
• Check that your user ID is whitelisted if you set up restrictions

Bot Responds Slowly

• Normal response time is 3–15 seconds depending on task complexity
• Very slow responses (30+ seconds) may indicate API issues on Anthropic’s end
• Check your internet connection or VPS connectivity

“Unauthorized” Errors

• Your bot token may be invalid or revoked — generate a new one from BotFather
• Make sure you copied the full token with no extra spaces

Bot Works But Stops After a Few Hours

• OpenClaw likely crashed or your session ended — use PM2 to keep it running persistently
• On a VPS: pm2 start openclaw -- start && pm2 save && pm2 startup

Pro Tips for Daily Use

• Pin your bot chat: In Telegram, long-press the bot chat and pin it for quick access
• Create a shortcut: Add the bot to your phone’s home screen for one-tap access
• Use voice messages: Many versions of OpenClaw support voice message transcription — speak your tasks instead of typing
• Send files directly: You can send documents, images, and files to the bot for processing
• Set custom notifications: Configure Telegram to give your bot a distinctive notification tone so you notice proactive messages

Next Steps

With Telegram connected, your OpenClaw agent is fully operational. Explore what it can do:

• Install Skills to add new capabilities
• Master all available commands
• Ideas for using OpenClaw productively

🛒 Recommended: Automation Business Book | Productivity Desk Mat

Want to see what OpenClaw can really do? Check out this wild project building AI agents with physical bodies →