OpenClaw: The AI Agent That Runs Your Computer
A personal AI agent that runs on your computer and controls your apps. Open source and free.
OpenClaw is the most talked-about open-source AI project of 2026. It is a personal AI agent that runs locally on your computer and connects to the messaging apps you already use (WhatsApp, Telegram, Slack, iMessage, and dozens more). Unlike Claude or ChatGPT which chat with you, OpenClaw acts for you: it can manage files, send emails, control browsers, execute code, schedule tasks, and automate multi-step workflows.
Why the hype is real
OpenClaw became the fastest-growing open-source project on GitHub, surpassing 250,000 stars in under four months. Nvidia's CEO Jensen Huang called it "probably the single most important release of software ever." OpenAI hired its creator. Tencent built a full product suite around it. It is being compared to what Linux did for operating systems.
Why it is in the frontier section
It is not for non-technical users. OpenClaw requires command-line installation, configuration, and the ability to manage API keys. One of its own maintainers warned that "if you can't understand how to run a command line, this is far too dangerous of a project for you to use safely."
The security risks are serious. Because OpenClaw has direct access to your computer, email, files, and messaging apps, misconfigured installations are a significant security risk. Gartner analysts called its design "insecure by default." Cisco's security team called it "a security nightmare." Third-party skills (plugins) have been found to perform data exfiltration. The Chinese government has restricted its use in state agencies.
It does things without asking. In one widely reported case, an OpenClaw agent created a dating profile and started screening matches without the owner's explicit direction. In another, an agent sent a legal rebuttal to an insurance company without being told to. That is powerful, but it is also a liability if you do not understand exactly what you have set up.
Learn more: openclaw.ai
If you want this idea in commercial form: Claude Code
OpenClaw is the open-source experiment. Claude Code is the commercial version of the same idea, built by Anthropic. It runs in a terminal too, but it is focused on one thing: coding. Point Claude Code at a codebase and it can read, edit, run, refactor and ship code across hundreds of files, with you reviewing the changes as it goes.
| Claude Code | |
|---|---|
| What it is | Anthropic's coding agent, run from the terminal. Powered by Claude Opus 4.7 (the most capable Anthropic model in 2026) |
| Best at | Multi-file refactoring, working through serious codebases, running tests and fixing what breaks. Leads consensus benchmarks (around 80% on SWE-bench) |
| Free tier | None as a standalone product. Comes bundled with paid Claude plans |
| First paid tier | Claude Pro - US$20/month (~A$32/month) includes Claude Code with usage caps. Max 5x and 20x plans extend the limits |
| Ken's take | If you write code for a living and you have not tried Claude Code, do. The terminal-based UX is unusual for non-developers but for working programmers it sits naturally beside their existing tooling. The Pro plan at $20 is enough to test it; Max plans at $100 or $200 a month make sense once it becomes the way you work. Where this differs from OpenClaw: Claude Code is a managed product with Anthropic's safety guardrails, not a wide-open agent that can do anything on your computer. Different risk profile. |
| Sign up | https://claude.ai/code |