All insights

Insight · January 31, 2026

CalwdBot? MoltBot? OpenClawd?

AI is moving fast, really fast, this week alone we had a new tool come out that has taken over the internet by storm. It started out as ClawdBot...

# OpenClaw: The Open Agent Platform Taking Over the Internet AI is moving fast, really fast. This week alone, a new tool emerged that has taken the internet by storm. It started out as ClawdBot, then got renamed to MoltBot. Anthropic sent a letter to the creator, Peter Steinberger, as the name sounded too similar to Claude. As of January 29th, 2026, Steinberger renamed it to [OpenClaw](https://openclaw.ai/blog/introducing-openclaw). OpenClaw is a free, open agent platform that runs on your machine and works from the chat apps you already use—WhatsApp, Telegram, Discord, Slack, Teams. Wherever you are, your AI assistant follows. Think Claude Code, but it can control your computer and acts as your 24/7 assistant. You can text it commands from your chat app, eliminating the need to have your computer with you to use Claude Code. You can use any LLM you'd like by simply adding your API keys: OpenAI, Google Gemini, and more. The current installation process still requires some technical knowledge, such as using a terminal, but it can easily be done by anyone. --- ## What Is OpenClaw? Peter Steinberger, who has built Apple tooling for years, created OpenClaw (originally ClawdBot) back in December 2025. The project goals were simple: create a fully self-hosted AI agent where your prompts and files never leave your hardware except when sent to an LLM API you configure. This was loved by developers who were tired of routing everything to the cloud. It also meant increased privacy (initially). [The repo currently](https://github.com/openclaw/openclaw) sits at 135,000 stars. --- ## How Does It Work? Before diving in, let's break down **daemon** and **gateway**. A **daemon** is just a program that runs in the background on your computer. No window, no interface—it just sits there waiting. Your computer already has a bunch of them. The thing that syncs your photos to iCloud? Daemon. The thing that makes Spotlight search fast? Daemon. They start when you boot up and quietly do their job until you shut down. A **gateway** is a bridge between two different systems. It takes input from one side, translates it, and passes it along to the other. OpenClaw combines both: it's a background process that keeps your messaging connections open and routes everything between your chat apps and the AI agent. Here's what happens when you send a message: 1. Your text comes in from WhatsApp, Telegram, Discord, iMessage, or the CLI. 2. The Gateway passes it to an embedded coding agent called Pi, which can run shell commands on your machine. 3. Pi sends your prompt to whatever model you've configured (Anthropic, OpenAI, Gemini, etc.). 4. The model responds with instructions, and Pi executes them locally. 5. Results get sent back through the Gateway to your chat. The Gateway also handles session management, response streaming, and a local web dashboard for configuration. It installs as a system service (launchd on macOS, systemd on Linux), which means it survives reboots and you don't have to think about it. --- ## Security Concerns OpenClaw is impressive and represents something all agentic developers have been trying to achieve. But looking at it from a security lens, it's concerning: - **Unrestricted system access**: OpenClaw runs shell commands, has read/write file capabilities, and can execute scripts on your machine without intervention. If configured this way, you've granted it access to do anything with your files. There have been many cases where Claude Code has deleted entire databases (mine included). Now imagine if that happened with one of your files—or if a file got sent to the wrong person. - **API key exposure**: It has already exposed many API keys and entire databases. Jamieson O'Reilly wrote an article on [hacking into ClawdBot](https://x.com/theonejvo/status/2015401219746128322) and retrieving API information from other users. - **Prompt injection vulnerabilities**: This is a major issue where threat actors can craft malicious prompts and cause unintended or unauthorized behavior. If you're thinking about installing OpenClaw, please read and implement the [security best practices](https://docs.openclaw.ai/gateway/security). --- ## Conclusion OpenClaw has the potential to transform the way you work and will mark the beginning of some amazing advances in agentic AI in the coming months. Hundreds of new AI startups will emerge, and hundreds will die off or pivot. What do you think is the future of agentic AI? Email me at [hashim@bytesavy.com](mailto:hashim@bytesavy.com) to talk more about this.
Skip to main content