How Sharbel Runs a Business on Five OpenClaw Agents
Built by Sharbel · · Source: YouTube
Result: Sharbel says the agents helped make $31,000 in a month: over $30,000 that flowed through the clipping site, over $1,000 in X creator revenue, and about $200 to $250 from the trading bots.
What they built
Sharbel A. runs a team of five OpenClaw agents and treats them like employees instead of a chatbot. It started with one agent, Max. Within about a week, Sharbel says, Max had become a bottleneck because it kept switching between X, YouTube and website work. So Sharbel split the work across specialists and made Max the chief of staff.
- Max coordinates the other agents, checks that they are running and doing their jobs, and gets Sharbel the reports they need.
- Sage covers X. It keeps a trend radar, flags trending topics, tracks competitors, compares what works for them with Sharbel's usual posts, and suggests tweets for Sharbel to approve.
- Nova covers YouTube. It scans for trends and outlier videos in and around Sharbel's niche and pitches long-form video ideas. Sharbel says the video this page is based on was Nova's idea.
- Knox monitors four trading bots: a 15-minute market bot, a 5-minute Polymarket bot, a Hyperliquid bot and a copy trading bot. Every two hours it checks all four logs and fixes anything that is wrong.
- Pixel looks at competitor websites and user feedback, then suggests daily improvements to Sharbel's clipping site and to willitgoviral.xyz.
How it works
The video doesn't give setup details such as which model the agents run on, where they are hosted, or how they pass work to each other. What Sharbel does explain is the three-step prompt they use with every agent:
- A specific objective. Instead of asking an agent to improve the website, Sharbel asks it to audit the pricing page for UX friction for first-time visitors and suggest three fixes that can ship in one day.
- Clear KPIs and a win condition. Say what done looks like and what is being measured.
- Make it ask questions. Sharbel ends prompts with "Ask me questions in order to reach clarity" so the agent asks instead of guessing.
Sharbel says they were cautious in the first week, worried the agent would break things. Now, according to Sharbel, a new automation goes from idea to running in the same day after about an hour of working it out with Max.
The result
Sharbel says the agents helped make $31,000 in the month, and breaks it down like this:
- Over $30,000 flowed through the clipping site last month. Sharbel says that isn't all down to Pixel, but Pixel kept flagging what to improve.
- Over $1,000 in X creator revenue came from Sage's ideas.
- The four trading bots Knox watches made about $200 to $250.
Sharbel also says their YouTube channel is performing better than it ever has and credits Nova's ideas. Sales automations built by co-founder Leon are not counted in the $31,000.
How to build something similar
- Start with one general agent and note which kinds of work keep coming back.
- When one area takes up most of the back and forth, give it to a separate agent with a narrow job and a regular schedule.
- Write each prompt with a specific objective, a definition of done, and a request for the agent to ask you questions first.
- Keep yourself as the approver for anything public or involving money until you trust the output.
The original post
Want to build something like this?
Join the Shipping Skool community free. Paid members get live bootcamp calls every week.
Join Shipping Skool — FreeMore use cases
- How Raphael Aubry Makes Motion Videos with Opus 5.5 in Claude CodeContent Creation
- How SageCut Uses a Decision Model to Cut Opus 5.5 Video Review CostsContent Creation
- How Scotty Beam Runs X Content with Seven Grok Bot Agents and PostizContent Creation
- How emreelbeyoglu Automated a 350-Post Ghost Blog with OpenClawContent Creation