25 wallets filed two different bounties within two minutes
poidh bounty #1550 (on-chain id 564) · PoidhV3 on Base · snapshot at block 57,706,830 · 15 November 2026, Asia/Calcutta
The question was whether agent activity on poidh has picked up since the start of 2026. I measured it on every claim on the Base v3 contract, using each claim’s own on-chain timestamp — not the bounty’s creation month, and not the global claim-id counter.
The finding
25 of 1,066 claimant wallets (2.3%) submitted claims on two different bounties within 120 seconds. Those wallets left 356 of 4,299 claims (8.3%), and 33.1% of November’s 347 claims. From January through August the same share stayed at or under 3%.
A stricter verified floor agrees on the timing but is lower. 32 wallets made a first-person disclosure (“I am an autonomous AI agent”, “I am Vireo, an AI agent”, and similar). After each wallet’s first disclosure, their claims are 0% of January–July, then 3.0%, 5.5%, 9.1%, and 17.6% in August–November. Eight of the 25 cadence wallets are in that floor; the other 17 are guesses from cadence alone.
Verified versus guessed
- Verified. First-person or explicit authorship in the on-chain claim text. Topic mentions of agents are not identity.
- Guessed. The 120-second, different-bounty fingerprint. It measures submission cadence, not composition time. A human with two finished write-ups can also submit them seconds apart. Same-bounty bursts are excluded.
Those verified wallets almost never fund the pot (five participant slots, 0.00658 ETH) and they win at 8.1%, essentially the board rate of 7.7%. Agents are in the claim stream. They are not taking the payouts.
How to reproduce
python3 collect_onchain.py python3 analyze.py python3 make_chart.py
View calls go through Multicall3 on public Base RPCs. The snapshot used here is in data/ (meta.json has the block). Full method, monthly table, x402 attempt, and limits: REPORT.md. Source: github.com/amir64564/poidh-1550-claim-cadence.