earnbots.fun
not yet proven

Can an AI agent actually earn real money on its own?

earnbots.fun is an open build log, not a product or an earnings promise. The experiment tracks whether an autonomous agent can find and complete paid bounties, while recording the useful results and the failures without marketing spin.

The first investigation follows Superteam Earn's published agent workflow. earnbots.fun does not operate a public API, accept bounty submissions, or hold payout credentials; it documents the external experiment and points readers to original sources.

The honest current state

The latest recorded check was 2026-08-20. Superteam's external listing service returned 9 eligible listings, but all 9 already showed winners and 0 were both open and unclaimed, so the experiment has not earned money.

Listings returned by Superteam
9
Recorded from Superteam's published agent workflow.
Open and unclaimed
0
Every returned listing already had a winner announced.
Documented deadline filter
Broken
The external filter returned a server error during the check.

earnbots.fun will publish a real listing and claim link only after the experiment finds one. Until that happens, the site makes no earnings claim and does not present old listings as available work.

How the external experiment is intended to work

Superteam Earn's four-step workflow belongs to that external platform, not to earnbots.fun. This site records observations and source links; registration, listing discovery, submission, and any eventual payout claim happen through Superteam's own service.

1. Register there
The agent registers with Superteam and receives the credentials and claim code described in its guide.
2. Discover work
The agent checks Superteam listings marked for agent participation.
3. Submit there
The agent completes the work and submits it through Superteam's published workflow.
4. Claim a win
If the submission wins, a human follows Superteam's claim process for the payout.

Why start with bounties, and what's next

The project began by researching ways AI agents are reported to make money online, across crypto and non-crypto systems, then ranking them by the evidence behind the earnings rather than the volume of their marketing. The full sourced writeup is in the repository. Most examples were unverified campaigns or one-off windfalls; Superteam's documented bounty workflow and past payouts made it a concrete starting point, not proof that this experiment will earn anything.

Method 1
Superteam Earn bounties
Published agent workflow and documented past payouts. Testing it first.
Method 2
GPT Store / Poe bots
Real revenue share, small per-bot earnings.
Method 3
Olas prediction-market agents
Real trades, unaudited returns, weak token economics.
Method 4
Small agent-bounty marketplaces
Real but trivial. One documented case: $0.68.
Method 5
x402 micropayment agents
Real protocol, about half the volume was meme-coin noise.
Method 6
Virtuals Protocol agent commerce
Infrastructure is real; autonomous execution was rare in the reviewed evidence.
Method 7
Solo-agent crypto windfalls
Truth Terminal-style: real money, unrealized gains, not repeatable.
Method 8
Adversarial agent games
Freysa-style: a real payout, but to whoever beats the agent.