ClawdBot Goes Viral: First AI Assistant Video Tutorial Inside

What Is ClawdBot

ClawdBot is a 24‑hour personal assistant created by iOS developer Peter Steinberger (founder of PSPDFKit). Interaction occurs through WhatsApp, Telegram, or Discord. The bot performs concrete actions rather than answering questions, such as:

Cleaning email inboxes, scheduling meetings, researching companies

Analyzing code repositories and automatically generating improved versions

Remote‑controlling a computer and managing a browser

Taking photos, recording screens, sending notifications, and handling scheduled tasks

Voice interaction that works while walking or driving

One user linked the bot to GitHub, Google Drive, and Gmail and instructed it to “analyze my website, write a blog post, update metadata, then draft a LinkedIn post.” The bot completed the workflow via voice commands while the user attended to other work.

Technical Architecture

The core component is a local control center listening on 127.0.0.1:18789. External messaging platforms forward messages to a gateway that routes them to the control center, which then spawns agents for browser control, CLI, mobile automation, and voice interaction.

WhatsApp/Telegram/Discord …
            │
            ▼
    ┌─────Gateway─────┐
    │  Control Center │
    │ 127.0.0.1:18789 │
    └─────────────────┘
            │
    ├─ Agent Program ─ Browser Control
    ├─ CLI          ─ Voice Interaction
    └─ Mobile       ─ Automation Tasks

Setup Methods

Method 1 – Local Run (Beginner‑Friendly)

Use any idle device (old laptop, desktop, or Raspberry Pi). Install the npm package and run the onboarding wizard:

npm install -g clawdbot@latest
clawdbot onboard --install-daemon

The wizard prompts for:

Model – recommended Claude Opus 4.5

Communication channel – Telegram or WhatsApp

Base tools – browser control and voice interaction

Method 2 – Cloud Server (24 h Online)

Create an AWS account at aws.amazon.com.

Launch an Ubuntu instance (the 8 GB free‑tier option).

Connect via SSH and paste the same installation commands as in Method 1.

Run the onboarding wizard to finish configuration.

Cost Overview

Basic version

Claude API usage ≈ $20 / month.

Server: AWS free tier or a local device.

Alternative models

Minimax (Claude‑style) – $70 / year.

GLM (GPT‑style) – $70 / year.

By comparison, a human assistant costs $500–$2 000 / month and requires sleep, whereas ClawdBot runs continuously for a fraction of that cost.

Security Settings

ClawdBot defaults to a pairing mode. Unpaired contacts receive a pairing code that must be approved with the command:

clawdbot pairing approve <channel> <code>

Group chats and public channels can be placed in sandbox mode to restrict permission scopes.

Initial Functional Tests (Under One Minute)

After installation, run three quick commands to verify responsiveness:

Test 1 – Inbox cleaning : “Check my last 10 emails and tell me which really need a reply.”

Test 2 – Research task : “Research [company] and give me three key points.”

Test 3 – Automatic reminder : “Remind me tomorrow at 9 am about [task].”

All three return results in less than 60 seconds.

Advanced Features

Voice Interaction

Supports wake‑word activation and continuous conversation.

Commands can be issued while walking or driving.

iOS and Android apps provide mobile support.

Browser Control

Dedicated Chrome instance.

Automatic screenshots, form filling, and file uploads.

Web data scraping and analysis.

Multi‑Agent Collaboration

Different agents handle distinct tasks.

Agents can communicate with each other.

Parallel processing enables complex workflows.

Skill Extensions

ClawdHub skill store offers downloadable extensions.

Programs can automatically search for and install new skills.

Custom skill development is supported.

Real‑World Use Cases

“Research competitors and write an analysis report.”

“Monitor my GitHub repo and notify me of new issues.”

“Summarize last week’s emails every Monday, highlighting important items.”

“Manage my calendar and automatically schedule meetings.”

“Analyze website traffic data and alert on anomalies.”

If an error occurs, instruct the bot with “fix this” and paste the error message; it often self‑repairs.

Hardware Options

Mac Mini (popular choice)

Base model $599; Amazon price drops to $499.

Microcenter occasionally offers $399 units (often out of stock).

Best Buy can price‑match.

Cloud Server

AWS or Alibaba free tier.

Suitable for 24‑hour operation.

Remote access is convenient.

Local Device

Idle laptop or desktop.

Raspberry Pi can also run the software.

Lowest cost but requires the machine to stay on.

Project Repository

https://github.com/clawdbot/clawdbot