# How I Use Clawdbot to Run My Business and Life 24/7 Source: https://www.youtube.com/watch?v=YRhGtHfs1Lw Recap page: https://rapidrecap.app/video/YRhGtHfs1Lw Generated: 2026-01-29T20:36:08.936+00:00 --- ## Quick Overview The creator demonstrates using Clawdbot, an AI assistant, to automate personal and business tasks by leveraging its ability to interact with various channels like Discord, Telegram, and external APIs, ultimately aiming for a highly productive, self-hosted AI ecosystem that processes data locally or via secure APIs, avoiding reliance on proprietary systems for sensitive data like bank transactions or health info. **Key Points:** - Clawdbot acts as a central AI assistant, connecting to channels like Discord, Telegram, Trello, and Watch via self-hosted Mac Studio to automate tasks. - The creator explicitly avoids connecting Clawdbot to sensitive data sources like bank transactions or health data, preferring self-hosting or using secure APIs for those functions. - The system uses custom HTML dashboards cast to Google Home devices to display information like recent dental work history, which was manually exported from banking/dental records. - The creator uses Clawdbot to manage personal tasks like finding specific YouTube playlists (e.g., children's songs) and scheduling actions like calling a phone number. - A key feature discussed is the ability to create custom agents (personas) with specific skills (e.g., Benji for coding, Dr. Cox for health) to handle different types of requests. - The creator recommends starting with Discord or Telegram integration for users unfamiliar with self-hosting, as it's less complex than other methods like self-hosting a full instance or using a VPS. - The creator is actively developing the system, noting that prompt injection attacks are a risk and that the AI will continue to learn and evolve over time. ![Screenshot at 00:37: The Clawdbot Architecture diagram illustrates the flow of data from user devices \(Phone, Macbook, Assistant, Watch\) through various Channels \(Discord, Telegram, Trello, Webhooks\) into the central 'Mac Studio' processing unit, which then interacts with Cloud services \(Kefir, Postgres, etc.\), Smart Home devices, and External APIs/Health Trackers.](https://ss.rapidrecap.app/screens/YRhGtHfs1Lw/00-00-37.jpg) **Context:** The video features a discussion between two individuals, one of whom (Kitze, the creator of Clawdbot) showcases his highly customized, self-hosted AI assistant setup called Clawdbot, which he uses to manage aspects of his personal and business life. Clawdbot is designed to integrate with numerous communication channels and external services, with a strong emphasis on data privacy and local processing where possible. The discussion revolves around the practical, everyday use cases and the underlying architecture of this advanced automation framework. ## Detailed Analysis The creator demonstrates Clawdbot, an AI assistant running locally on a Mac Studio, designed to handle automation across personal and professional domains. The architecture, shown in a diagram (00:37), involves user inputs from various channels (Discord, Telegram, Watch) feeding into the 'Mac Studio' core, which communicates with Cloud services and external APIs. The creator emphasizes extreme caution regarding data privacy, explicitly stating that he manually exports sensitive data like bank transactions (13:54) and health data, rather than allowing Clawdbot direct, unfiltered access to these sources. He uses custom HTML dashboards cast onto Google Home devices to visualize this sensitive data, as seen in the screenshot displaying dental history (23:38). The system supports custom agents with specialized skills, such as one persona dedicated to coding (23:34) and another for managing personal health data. The creator suggests users start with simpler integrations like Discord or Telegram before attempting complex self-hosted setups to avoid security pitfalls like prompt injection (13:19, 21:21). He highlights that the system is designed to learn from interactions, enabling it to handle increasingly complex, personalized tasks like managing smart home devices or filtering emails, leading to a highly efficient, personalized automation layer. ### Clawdbot Architecture Overview - Channels (Phone, Discord, Telegram, etc.) feed into 'Mac Studio' core logic which connects to Cloud services (Kefir, Postgres) and external APIs/Smart Home/Health Tracking - Data flow is complex, involving multiple input/output arrows. ### Demonstrated Use Cases - Displaying custom HTML dashboards on Google Home (23:38) showing dental history (13:15); managing specific tasks like checking flight status or handling emails (00:00); creating custom agents for coding (23:34) and health monitoring (04:44). ### Security & Privacy Focus - Creator manually exports sensitive data (bank, health) rather than allowing direct AI access (13:54); warns against prompt injection (14:18); prefers self-hosting/local AI to avoid exposing data to cloud services. ### Tinkering & Future - Creator mentions having a 'Tinkerer Club' (29:58) for like-minded people; notes that the system is constantly learning and iterating (14:55); mentions the potential for a massive split between consumer AI and tinkerer AI (23:13). ![Screenshot at 00:00: The initial screen showing the Moltbot landing page with positive testimonials.](https://ss.rapidrecap.app/screens/YRhGtHfs1Lw/00-00-00.jpg) ![Screenshot at 00:37: The detailed diagram illustrating the Clawdbot Architecture, showing inputs from user devices, channels, the central processing unit, and external integrations.](https://ss.rapidrecap.app/screens/YRhGtHfs1Lw/00-00-37.jpg) ![Screenshot at 04:45: A view of the creator's complex Discord server structure, organized into numerous channels for different projects and tasks.](https://ss.rapidrecap.app/screens/YRhGtHfs1Lw/00-04-45.jpg) ![Screenshot at 23:38: A screenshot of the custom HTML dashboard displaying dental records on a Google Nest Hub, demonstrating a personalized data visualization capability.](https://ss.rapidrecap.app/screens/YRhGtHfs1Lw/00-23-38.jpg) ![Screenshot at 27:14: The 'Use Prompt' interface within Spellbook, showing how variables are populated to generate complex code snippets or instructions for a new Swift app.](https://ss.rapidrecap.app/screens/YRhGtHfs1Lw/00-27-14.jpg)