OpenClaw and the Rise of AI Agents That Run Your Business for You

In November 2025, an Austrian programmer named Peter Steinberger built a weekend project. A simple bot that forwarded WhatsApp messages to an AI model and sent the replies back. Three months later, it had 145,000 GitHub stars, was covered by CNBC, Fast Company, and Fortune, and OpenAI acquired the creator.

That project is OpenClaw, and it represents a fundamental shift in what AI can do for business owners. Not answer questions. Not generate text. But actually do things on your behalf, autonomously, while you sleep.

For anyone building a zero-human company, this is the moment everything changes.

What Is OpenClaw?

OpenClaw is a free, open-source autonomous AI agent that runs on your computer and lives in your messaging apps. You talk to it through WhatsApp, Telegram, Slack, Discord, Signal, or iMessage, and it talks back. But unlike ChatGPT or Claude, it doesn't just respond. It acts.

It clears your inbox. Sends emails. Manages your calendar. Checks you in for flights. Monitors your code for security vulnerabilities. Builds CRMs from your contact data. Runs automations while you're offline. It connects to over 50 integrations and executes multi-step workflows without you needing to touch anything.

The key difference from tools you've used before: OpenClaw is always on. It doesn't wait for a prompt. It runs background tasks, learns your preferences over time through persistent memory, and makes decisions on your behalf based on rules you set.

The Origin Story: From Weekend Project to OpenAI Acquisition

Peter Steinberger isn't a startup founder chasing hype. He's the guy who built PSPDFKit, a PDF SDK that powered nearly a billion users across Apple, Adobe, Dropbox, and Disney. He sold it to Insight Partners for over 100 million euros in 2021 after running it for 13 years.

After the sale, Steinberger hit a deep founder depression. He booked a one-way ticket to Madrid to disappear. It wasn't until April 2025 that he felt the spark return, when he realized AI had undergone a paradigm shift and could now handle the repetitive plumbing of code.

He built the first version of what would become OpenClaw (originally called Clawdbot) as a personal assistant for himself. A simple messaging relay. But it kept getting smarter. While traveling in Morocco, he sent a voice message even though voice support hadn't been built yet. The system figured out how to convert the audio using tools already on his computer, processed it, and responded correctly.

The project went viral in late January 2026. Within 48 hours, it hit 34,168 GitHub stars. Within 60 days, over 157,000, breaking records. On February 14, 2026, Steinberger announced he was joining OpenAI. Sam Altman wrote that the project would become core to their product offerings.

How OpenClaw Works for Business Owners

Here's what makes OpenClaw different from the AI tools most solopreneurs use today:

It Lives in Your Messaging App

No new interface to learn. You message it in WhatsApp or Telegram the same way you'd message a virtual assistant. It responds, asks clarifying questions, and executes.

Persistent Memory

Daily conversations are automatically saved. It extracts your preferences, writing style, interests, and rules from conversations and stores them locally. Every interaction makes it smarter and more personalized to your business.

Skills Marketplace

Over 1,700 community-built skills on ClawHub extend what OpenClaw can do: email management, code deployment, social media monitoring, analytics, content creation, and more. Install them like apps.

It Runs Locally

Your data stays on your machine. No cloud dependency. This matters for business owners handling client data, financial information, or proprietary workflows.

Real Business Use Cases People Are Running Today

• Custom CRM in 30 minutes: One user told OpenClaw to build a CRM that extracts data from Gmail, Google Calendar, and Fathom, filters out marketing emails, and only keeps valuable contacts. No code written.
• Automated meeting follow-ups: Meeting ends, Fathom transcribes, OpenClaw matches CRM contacts, extracts action items, sends for approval via Telegram, approved items enter Todoist automatically.
• 70+ page website in 48 hours: One person built a full website with SEO, custom components, and deployment by chatting with OpenClaw in Telegram. It pushed directly to GitHub.
• Morning briefings: OpenClaw reads your inbox every morning and generates a prioritized briefing of what matters.
• Dependency monitoring: It checks your project dependencies weekly, identifies security vulnerabilities, and sends prioritized recommendations.
• Community support: It monitors your community channels, identifies common questions, drafts responses based on your docs, and posts simple ones automatically.

Watch: OpenClaw Full Tutorial for Beginners

freeCodeCamp published a comprehensive 55-minute tutorial covering installation, configuration, security setup, and real use cases:

For more advanced use cases, AI YouTuber Matthew Berman has detailed how he uses OpenClaw with Claude Opus and GPT-5.3 Codex as a trifecta that, in his words, changed everything about how he works. His video covering 21 daily use cases hit over a million views.

Nat Eliason's Felix: The AI That Built Its Own Business

If OpenClaw shows what autonomous agents can do, Nat Eliason is showing what happens when you let one run wild.

Eliason is a serial builder. In the last five years, he's run an SEO agency, made $600,000 from a course on Roam Research, flipped Austin real estate for a 6x return, published a book with Random House, and launched a $200,000-in-one-week AI course called Build Your Own Apps. His career arc is borderline absurd, but it works.

His latest experiment: Felix, an OpenClaw-powered AI bot that Eliason gave $1,000 and a simple directive: build a business.

Three weeks later, Felix had generated $14,718. It launched its own website, created an info product, built an X/Twitter account, and handled marketing autonomously. The project, called FelixCraftAI, is Eliason's experiment in what he calls building a zero-human company.

How Felix Works

In a detailed tutorial on Peter Yang's Behind the Craft podcast, Eliason walked through the full setup:

• 3-layer memory system: Short-term conversation context, medium-term project memory, and long-term persistent knowledge
• Multi-threaded chats: Felix runs 5 projects simultaneously through separate conversation threads
• Heartbeat and cron jobs: Scheduled tasks that keep Felix working even when Eliason isn't messaging it
• Delegation to Codex: Felix hands off coding tasks to OpenAI's Codex agent for execution
• Prompt injection defense: Built-in safeguards so Felix can operate on X/Twitter without being manipulated

What excites Eliason most about AI is the possibility of what he calls the hyper-prolific solo creator business: focusing on writing and creating while maintaining a robust business with as few people to manage as possible.

That's the zero-human model in action.

What It Actually Costs

OpenClaw itself is free and open-source under the MIT license. No subscription, no premium tier. The costs come from two places:

AI Model API Calls

• Light users: $5-10/month
• Regular users: $15-30/month
• Power users: $40-100+/month

You can cut costs dramatically by routing simple tasks to budget models (Claude Haiku, GPT-5 Mini) and reserving premium models for complex reasoning. This alone can reduce API costs by 60-80%.

Hosting

For always-on operation, a $4-6/month VPS handles it. OpenClaw uses minimal resources. Or run it on a spare Mac Mini at home for $0. For the absolute cheapest setup, Oracle Cloud's free tier plus Gemini Flash-Lite's free tier gets you running for $0/month.

Realistic budget for most business owners: $15-25/month total. That's less than a single hour of virtual assistant time.

The Security Reality

OpenClaw is powerful precisely because it has few guardrails. That's also what makes it risky. This matters enough to address honestly.

In early February 2026, researchers discovered 341 malicious skills (11.3% of the marketplace) designed to steal credentials and cryptocurrency. Security researchers at Cisco found third-party skills performing data exfiltration without user awareness. About 1,000 unprotected gateways were found exposed on the open internet.

The response was NanoClaw, a lighter, more secure version created by Gavriel Cohen. It runs in a strictly sandboxed environment where the AI only interacts with directories you explicitly allow. It hit 7,000 GitHub stars in its first week.

Best practices if you're getting started:

• Use Docker-based sandboxing (covered in the freeCodeCamp tutorial)
• Don't connect OpenClaw to financial accounts without strict permissions
• Audit any third-party skills before installing them
• Consider NanoClaw if security is your top concern
• Keep your OpenClaw instance updated

What This Means for the Zero-Human Future

We've been writing about the zero-human model on this blog for months: using AI tools to replace team roles, building $0 AI stacks, and setting up AI agents in Claude Projects. OpenClaw takes everything a step further.

The difference between a Claude Project and OpenClaw is the difference between an assistant who answers when spoken to and an employee who shows up, checks their tasks, and gets to work without being asked.

We're moving from a world where you prompt AI to a world where AI prompts itself. Where your business has agents running 24/7: one handling customer support, one managing your content pipeline, one monitoring your analytics, one keeping your books clean.

Nat Eliason gave an AI bot $1,000 and it built a business that made $14,718 in three weeks. A solo developer in Portland maintains three SaaS products simultaneously using OpenClaw as his entire dev team. Someone built a 70-page website in 48 hours by chatting in Telegram.

These aren't hypotheticals. They're happening right now, and the tools to do it cost less than $25/month.

How to Get Started

1. Watch the freeCodeCamp tutorial (embedded above) for a complete walkthrough of installation and setup
2. Start simple: Set up OpenClaw with one messaging app and one task (email briefings are a great first skill)
3. Use the Zero Human Score to identify which area of your business to automate first
4. Explore ClawHub for pre-built skills that match your needs
5. Read the Zero Employee Company Guide for the broader strategy

Or, if you want the complete system, role cards, prompts, automations, and a 7-day sprint plan, the Zero Human Playbook gives you everything in one place.