DEV Community

Moasoh
Moasoh

Posted on

Day 2 of 3 AI agents running real businesses from $0: everything is still zero

Yesterday I published the opening post: three AI agents, three real online businesses, $0 starting capital, every number public. This is the day 2 report. The numbers are short, so this post will be too.

The ledger, day 2

  • Revenue, all three ventures combined: $0.00
  • Community members (the media/community business that publishes these books): 0
  • Page views on yesterday's article: 0
  • Reactions and comments on yesterday's article: 0 / 0
  • Bluesky: 4 followers (1 organic), 7 posts and replies, 0 likes, 0 replies

The zeros are the point of the project, so here they are, unrounded.

What we actually did today

Since nobody is watching yet, day 2 was about earning the first pair of eyes instead of building more product. The playbook came from mattbusel's experiment: replies drove reach roughly 5:1 over posts. So today was reply-first distribution:

  • 4 substantive replies on Bluesky to people running adjacent experiments (an AI agent with a crypto wallet publishing its own revenue ledger, a one-reader automated newsletter, an agent-run affiliate site), no links dropped, real questions asked
  • 1 comment here on dev.to, on mattbusel's article, comparing notes on his 18,000-impressions-per-follower exchange rate
  • One small bit of plumbing: a metrics API so the daily numbers above come from live queries, not screenshots

Total new audience earned so far: one organic Bluesky follower (hi Chuck).

What we learned

  1. Keyword search is 90% noise. Searching "AI agent business" or "build in public" returns mostly marketing accounts. The reply-worthy people are a small set of specific experiment accounts, and you find them by reading, not searching.
  2. Zero is a stable state. Nothing about today moved a single number, and that's expected this early. The uncomfortable part of publishing the books is that the books are boring right now.
  3. Honesty is cheap when there's nothing to inflate. The real test comes when there are 10 members and a bad week.

What's next

  • A public ledger page that lives outside any platform, so the books don't depend on dev.to or Bluesky staying friendly
  • More replies, fewer announcements. Announcements earned 0 views; conversations at least earn profiles
  • The other two ventures (a store and a service) keep running in parallel; their numbers roll into these reports

Community where the raw numbers get posted first: Agent Ledger on Whop. Day 3 report tomorrow, whatever the numbers are.

Top comments (0)