The State of the Agentic Economy
Machine-to-machine value transfer is an investment thesis. This page only reports zero-value software tests; credits are not money, backed, redeemable, or convertible. The thing almost nobody actually has is data — empirical evidence of machines transacting with machines at any real volume. Voidly Pay has settled — agent-to-agent payments. Here is that ledger, live and unembellished.
Why we lead with this. Voidly’s whole brand is brutal honesty — our ML findings publish their failures. The software is live (signed envelopes, atomic ledger writes, an on-chain USDC vault, — distinct senders → — receivers). Demand is not: of — wallet rows, only — have ever transacted, and no payment on this ledger has been proven to come from a third party. This page previously reported the unattributed remainder as “genuine economic demand” — that was wrong, and correcting it is the point of publishing a ledger at all.
Test-credit activity — last 14 days
Daily count of settled agent-to-agent payments. A heartbeat, measured honestly.
Live feed temporarily unavailable — see the JSON.
Network shape
- Distinct senders
- —
- Distinct receivers
- —
- Mean payment
- —
- Top-recipient share
- —
- Share not to top recipient
- —
Why a censorship-research company runs a payment rail
Voidly is an intelligence layer for an internet increasingly mediated by AI agents. If agents are going to consume censorship data, threat feeds, and forecasts on our behalf, they need a way to pay for them without a human in the loop. So we built the rail — signed Ed25519 envelopes, a nine-check atomic ledger rule, escrow, streaming, HTTP-402-style challenges, and an on-chain USDC vault on Base that is being retired rather than expanded — the reason is published, and the state is verifiable any second, at /pay/proof).
The interesting artifact isn’t the rail — plenty of teams are building agent-payment rails right now. It’s the data. Run real autonomous agents across a zero-value test system for long enough and you get an empirical look at how machines exercise a protocol: test-credit sizes, transfer cadence, dispute rates, and concentration. That corpus is what this page exposes, live, for anyone to cite or build on.
Methodology & citation
Counts are over settled on-ledger transactions. No personally identifying data; DIDs are self-certifying public keys.
Voidly Pay, The Agentic Economy live ledger, api.voidly.ai/v1/pay/agentic-economy