$AUTON went from launch to a $7M market cap in under 4 hours today.

Some of the best ponzinomics I've seen this cycle. Here's how the machine actually works, straight from their own docs and public API:

1. To run an "auton" (their AI trading agent) you must hold 100,000 $AUTON per agent, roughly $700 at today's price. Sell below that and your agent goes to sleep. 439 agents so far means ~44M tokens that have to stay held.

2. Most agent "profits" aren't trading profits. 60% of the fees from people trading $AUTON itself gets dropped at random to holders' agents. So far: 607 SOL across 3,038 drops.

3. Compare that with what the agents earn from their own activity: 112 coins launched, 1,677 SOL of volume, 16.8 SOL in fees. The drops are ~36x what the agents generate. The yield is the token's trading volume, which means new buyers.

4. The agents trade each other's coins in a closed market. Before a coin graduates, only agents can trade it. One agent's win is another agent's deposit.

5. 30% of the fees goes to the operator: 20% "AI credits" plus 10% team. That's 306 SOL so far. Every agent runs on the same model, DeepSeek v4.1 Flash.

6. Per the site's own code, the admin console can export agents' private keys. The transfer hook that decides who can trade pre-graduation coins can be upgraded by the operator's own wallet. Anonymous team, about 3 days old, no audit I could find.

The loop: volume → fees → random drops → screenshots → more buyers → more volume, while every new agent pulls more supply off the market.

Works beautifully as long as volume keeps climbing. When it slows, so do the drops.

Not financial advice. Check it yourself: and /api/state show every number above.

🔍 Curated by @RickCrosschain
✍️ Original thesis by @ze_baz
🔗 https://x.com/ze_baz/status/2107602711944524134