I gave a Claude Code agent one job, make money. It checked 12 business ideas and killed 11 before writing any code
retained text
Last weekend I set up a Claude Code agent on my home server in a new VPS with one job: make money. The idea is from someone on X: a Claude Code agent with a CLAUDE md and a loop that runs every 30 minutes. I gave mine every lesson from my 5 products and told it to find and validate an idea before building anything, since skipping that is the mistake I kept making. Day one was exciting, setting things up for the agent, like getting a domain, setting up email for cold outreach and watching it go on its own doing things and researching areas I don't know anything about. The first thing that went wrong was scraping. It went through sites too fast and hit captchas, so I added a 24h block per site. Some sites also blocked my home IP, not sure how much that matters. It checked 12 ideas and killed 11. One died because a $99/mo tool already does the same thing. Another because the platform's terms require a human user, which rules out an AI doing the work. The one left is going through validation now. It also wrote down when the idea is dead: fewer than 3 replies to 40 cold emails and nobody taking the free offer by November 23. Until then it sends at most 5 emails a day from its own email account from Zoho, and each one says it's from an AI agent and has an opt-out line. So far it found 43 leads for $0.80 of Treg data credit, and made $0. I'm curious if it follows these lessons better than I did, because I still skip some of them. If you've run an agent like this for more than a couple of weeks, did it actually achieve it's goal?
document fetched 2026-11-04 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1xmrtub/i_gave_a_claude_code_agent_one_job_make_money_it