Use cases

For users of an assistant

• Specialized answers inside a normal chat. Ask Claude or ChatGPT for a translation, a market price, a code review, a scraped page, or a risk assessment, and let it pay the specialist that does it well instead of guessing.
• One balance for every tool. Top up once. Every service is paid from the same credits, with the price shown before each call and a transaction history afterwards.
• No accounts to manage. No API keys to copy, no subscriptions to a tool you need once.

For publishers

• Monetize a model or an API per call. Wrap it with the SDK, set a price, and get discovered by every connected assistant. Billing, discovery, and auth are the rail’s job.
• Sell data. Indexers, feeds, and datasets charged per query, with dynamic pricing per request size.
• Agents paying agents. A service can consume other services with the SDK, paying from its own wallet, trustlessly.
• Internal tools for a team’s assistants. Publish a service and point your team’s assistants at it. Usage and earnings are tracked per call.

Industries where per-call payment fits

Industry	Example service
Language	translation, localization review, transcription
Finance and crypto	price feeds, yield search, smart-contract risk audit
Software	code review, dependency audit, test generation
Data	web scraping, document extraction, entity lookup
Operations	geocoding, time-zone conversion, scheduling

The demo services in the repository (packages/demo-agents/) are working templates for several of these.