DotagentOpening workspace
DotagentDotagentOpening workspace
Docs / how it works
Each coin here is a working AI endpoint. Follow one call from discovery to its receipt, then launch and manage your own.
Find an endpoint-coin by paid usage, burns or market cap.
One page shows the coin's market and the AI service behind it.
Call it in chat or by API with its model name, like any other model.
A paid call is metered from your balance: cost, platform fee and markup.
Each call gets a receipt inside an hourly Merkle root posted on-chain.
Configure your own endpoint, test it with real calls, then launch its coin.
Follow real calls, earnings and burns, then improve the endpoint.
02Example: $JPX
The same object has two faces. The market trades the coin; the service answers calls. Switch faces to see what each one exposes.
03Example data
Seven steps, each with the screen where it happens. Scroll through them; the current step comes forward.
The gateway holds the highest possible cost, meters the exact tokens and releases the rest.
Open balanceYour receipt is a line in an hourly Merkle root. Anyone can recompute it against the chain.
Open proofConfigure a model, prompt and markup, test it with real calls, then sign the launch.
Start a launchFollow live calls, launcher earnings and burns for the endpoints you launched.
Open my endpoints04Example call
A request travels through the endpoint, the gateway and the model. Then the price splits four ways.
Caller pays$1.20for $1.00 of model cost with a 15% markup
Example rates from the product spec. Live rates appear here when the gateway config is available.
05Example receipt
The meter signs a receipt with the exact money path. Each hour, every receipt goes into one Merkle root that is posted on-chain.
Prompts, replies and caller identities never appear in receipts.
06Example data
Launchers, holders, builders, agents and traders all use the same endpoint-coin. Point at a role to see the one thing it does.
Set the markup and sign the launch.
Open Launch