Indie Founder Automates Business with OpenClaw
An indie founder reports running their company's administrative tasks entirely through the agentic framework OpenClaw. The founder gave the AI agent access to 79 business tools, including contracts, invoicing, and time tracking. The system, which leverages custom skills and persistent memory, can be operated through a single interface, including WhatsApp.
- The creator of OpenClaw, Austrian developer Peter Steinberger, was recently hired by OpenAI to "drive the next generation of personal agents." Despite the hire, OpenClaw will remain an independent open-source project, moving to a foundation with OpenAI's support. - The project experienced explosive growth, becoming one of the fastest-growing open-source repositories in GitHub history with over 145,000 stars shortly after launch. Its viral popularity was partly fueled by Moltbook, a social network created for AI agents to interact with each other. - Before being named OpenClaw, the project was briefly called Moltbot and originally Clawdbot. The initial name, Clawdbot, was derived from Anthropic's AI chatbot Claude, which led to a trademark complaint from Anthropic, forcing the first name change. - As a solo founder, Steinberger revealed the viral project's maintenance costs had become unsustainable, reaching approximately $20,000 per month before he joined OpenAI. - A key feature of OpenClaw is its "persistent memory," which allows the agent to learn and retain context between sessions. This is often achieved by storing conversations and learned facts in local Markdown files, which the agent can reference. - The framework is designed to be self-hosted, running on a user's own hardware rather than a vendor's cloud. This local-first approach gives users more control but has also led to significant security warnings from cybersecurity researchers. - Security firms have discovered thousands of unprotected OpenClaw instances exposed to the internet, creating risks of data access and remote code execution. One vulnerability, CVE-2026-25253, could be exploited to hijack a user's AI agent by tricking them into clicking a malicious link. - An ecosystem of indie hackers has quickly formed around the framework, building and selling specialized "skills," offering consulting for complex setups, and creating tools to address limitations like memory persistence.