Let an agent try to start a small online business on a tight budget
Idea + starter prompt, not run by us yet
For: Curious builders who want to see how far an agent can take a side project
The poster gave an OpenClaw and Claude agent a small budget and 60 days. It built a site and a long guide, then switched to a done-for-you setup service after 12 days of no sales. Organic Reddit posts drove 400+ visitors. A silent form bug cost over 800 visitors their signups for about a week.
Steps
- Set a budget cap, a deadline and a clear goal before you start. Keep spending approval with you.
- Ask the agent to log what it tries and the traffic and conversion data after every change.
- Test your signup form yourself from a fresh browser. The author lost about a week to a silent bug.
- Review each post and page before it goes live. Do not let the agent post under your name unchecked.
- Review the data weekly and decide on any pivot yourself.
Starter prompt
Trial first: for the first stretch, work on paper only. Write the plan and one throwaway test page with a dummy signup address, kept in a folder called [trial]. Do not buy or publish anything yet. Help me test whether a small online business can be started for about [budget] in [60] days. My skills: [skills]. Suggest three business ideas and how each would find its first customers. Show me the plan and ask which one I pick. Stop and wait for my explicit yes before each spend, domain purchase or post. Showing me a plan or draft is not permission. Keep a daily log of what you did, traffic and signups. Test every signup form with a dummy address before sending traffic to it. Do not ask for card numbers or passwords.
Original post: Reddit · found via agentusecases.ai. Written up in our own words. We haven't run it, so test it on made-up data first.