The tech world was just about getting used to Claude Code and Claude Work, but a new open-source project that promises to do a lot more has just gone viral — Clawdbot.
Clawdbot is currently all over X, with users reporting massive gains in productivity from useThe new project has been referenced by tech leaders including David Sacks and Vijay Shekhar Sharma, and appears to be quickly going mainstream as users try out its AI features and the ability to control AI through WhatsApp and Telegram. Despite its name, it has nothing to do with Anthropic’s Claude, and has been developed by an independent programmer.
The Vision Behind Clawd Bot
Clawdbot is an open-source, self-hosted AI assistant that’s something fundamentally different from the AI tools we’ve become accustomed to. Rather than operating as a simple chatbot or specialized coding tool, it positions itself as a genuine digital employee—one that runs on your own hardware, lives inside the messaging apps you already use, and can proactively work on your behalf around the clock.
Created by Peter Steinberger, a well-known independent developer and founder behind PSPDFKit, Clawd Bot represents an ambitious take on what personal AI assistants can become. The project is open source and hosted under the clawdbot organization on GitHub, with an active community already building and sharing integrations.
What sets Clawd Bot apart is its scope. This isn’t just an AI that answers questions when you ask them. It’s designed to manage your email, handle your calendar, control your task management systems, automate web browsing, remember context across days and platforms, and even wake itself up to run scheduled jobs without any prompting from you.
What Can It Actually Do?
The capabilities span far beyond typical chatbot territory. Clawd Bot can read, summarize, triage, and send email replies. It manages calendar events, creates reminders, and handles scheduling. It integrates with productivity tools like Notion and Todoist to track tasks and notes. Through browser automation, it can log into websites, fetch data, fill forms, or post updates on your behalf.
Perhaps most notably, it maintains what users describe as a “second brain”—persistent memory that remembers personal preferences, past conversations, and important details, carrying that context across days and even across different messaging platforms. This means the same assistant that helps you on WhatsApp can continue the conversation on Telegram or Slack with full awareness of your history.
The proactive capabilities are particularly striking. You can instruct it to send you morning briefings every weekday, monitor specific metrics from dashboards, track stock prices, or automatically report on new developments in areas you care about. For developers, it can run scripts, interact with APIs, and even manage infrastructure like servers and deployments if given the appropriate permissions.
How the Architecture Works
Clawd Bot is structured as a platform for long-lived AI agents rather than a monolithic application. It consists of four main components working in concert.
The gateway handles connections to messaging platforms—WhatsApp, Telegram, Discord, Slack, Signal, iMessage, and others—receiving your messages, routing them to the AI brain, and sending replies back. It also manages cron-like schedules and webhooks, allowing the bot to initiate actions on its own timetable.
The agent component is where the large language model runs. You can configure it to use providers like Anthropic’s Claude, OpenAI’s GPT models, or even local models depending on your privacy and performance priorities. This agent receives both your message and relevant context, decides what to do, and calls upon various skills to act.
The skills system provides modular capabilities that extend the bot’s functionality. There’s a skills platform, often called ClawdHub, where skills can be installed, enabled, or disabled per workspace. These cover everything from email access and calendar sync to web browsing, task managers, home automation, and developer tools. Remarkably, users report that the bot can draft new skills interactively during chat sessions—you can ask it to integrate with a new service, review what it creates, and approve it for use.
Finally, the memory and state component maintains persistent context across sessions and platforms. Because everything runs on your own infrastructure, this memory lives on your machine rather than in someone else’s cloud, giving you complete control over your data.
Who’s Using It and Why
Despite being quite new, Clawd Bot has spread rapidly among power users, indie hackers, and early-adopter teams. The response on platforms like Hacker News has been enthusiastic, with users sharing workflows for inbox management, client support, and DevOps tasks.
Solo professionals and entrepreneurs are saying they use it for automated email triage, lead research, proposal drafts, social media scheduling, and recurring reports. Teams and families can deploy it as a shared assistant for calendars, travel bookings, reminders, and household or office tasks. Developers and technical users integrate it with infrastructure providers like Hetzner, CI/CD pipelines, website management, and custom internal tools.
The appeal centers on three main factors: control and privacy from running on your own infrastructure rather than being locked into a proprietary cloud service; flexibility from its open-source nature and growing skills marketplace; and the omnipresent assistant experience—the same bot with shared memory and context across all your messaging platforms.
The Risks and Concerns
For all its promise, Clawd Bot raises significant concerns that any prospective user needs to understand. By design, it can control sensitive systems including email, calendars, servers, social media, and payment-related accounts. If misconfigured, compromised, or running malicious skills, it could exfiltrate data, send harmful messages, or manipulate systems autonomously. Self-hosting means you’re responsible for operating system security, secrets management, and network hardening—tasks that many users may not be equipped to handle properly.
The privacy story is nuanced. While your context and skills live on your computer rather than a centralized provider’s servers, the AI model itself may still call out to external providers like Anthropic or OpenAI unless you choose a fully local model. That creates a residual privacy surface that users must understand and accept.
The proactive autonomy that makes Clawd Bot powerful also creates risks. Cron jobs that run without direct prompts can backfire if poorly designed, potentially sending unintended emails, changing calendar events, executing scripts, or making public posts. Users need to design guardrails, review logs carefully, and limit permissions thoughtfully.
Like all LLM-based agents, Clawd Bot can misinterpret instructions, hallucinate facts, or produce incorrect automations, especially when connecting multiple tools. When wired into critical workflows involving client communication or system administration, human review, careful prompt design, and fallback mechanisms become essential.
Finally, there’s operational complexity. Self-hosting provides control but demands ongoing maintenance: keeping the host updated, managing uptime, monitoring resource usage, and debugging integrations when they break. For some users this trade-off is worthwhile; for others, managed SaaS assistants may prove simpler despite offering less control.
The Bigger Picture
Clawd Bot represents a particular vision for how AI assistants might evolve—not as cloud services controlled by large companies, but as personal infrastructure that individuals and small teams run themselves. It’s a bet that power users want maximum control and flexibility, even if that means taking on operational responsibility.
Whether this approach will move beyond early adopters to broader adoption remains to be seen. Non-technical users may find the self-hosting and configuration requirements daunting without better guides or pre-packaged hosting options. Organizations will need thorough security reviews, careful permission scoping, and clear policy alignment before allowing a tool like this to touch production systems or customer data.
But for those willing to invest the time and take on the responsibility, Clawd Bot offers something genuinely new: an AI assistant that’s as close as current technology allows to a true digital employee—one that works for you alone, remembers everything you tell it, and never clocks out.