09 / 19PROPOSEDA design; nothing here is deployed
Service intents
ProposedPROPOSED. This chapter describes a design. Nothing in it is deployed, approved or running. Charging is off, TAON is not deployed, no TAO reserve exists and no audit has been done.
Bittensor is becoming operable by software at the chain layer, while the services built on subnets stay fragmented across each subnet’s own API. A service intent would let a program discover a route, preview what it would do and cost, and run it under a spending policy.
Bittensor’s own documentation describes chain operations as discoverable intents with machine-readable schemas, a plan or dry-run step before execution, spending and subnet limits, and machine-readable errors. Those cover chain transactions. The documentation read for this concept did not show an intent that pays for a service or returns a receipt for one (read through a summarising fetcher; to be checked against the raw pages before anyone relies on it).
| Part | What it would do | Today |
|---|---|---|
| Discover | List capabilities and eligible routes with their input schema and state. | Route state is readable at GET /api/intelligence VERIFIED |
| Preview | Return the selected route, an estimated cost and a latency range, without calling a provider. | PROPOSED |
| Policy | Let the caller set a spend limit per request or period, and the subnets allowed. | Only TAONIC’s own daily provider-spend ceiling exists VERIFIED |
| Execute | Run the previewed route and return the result. | Ask runs a request VERIFIED |
| Errors | Return a stable code and a fixed sentence for every failure. | Fixed failure sentences exist VERIFIED |
| Receipt | Return a record the caller keeps. | TAONIC writes it; it is not signed by the provider VERIFIED |
- Callers who want policy-bound calls across several subnets. This is unmeasured.
- Enough supply: only a handful of subnets sell a priced API today, so a neutral layer over them may be a feature rather than a market.
- Providers that sign or attest their results, or the receipt carries no evidence beyond TAONIC’s own book.
The opportunity is classed promising but unproven. No claim is made that TAONIC is first.
Every state on this page matches Status. Where they ever differ, Status is right and this page is a defect.