Technology

A substrate. Our model. Your memory. A rail.

Rivier is four layers working as one. A decentralized inference network underneath, so no single operator can take it down. Our own orchestration model on top, picking the right intelligence for each request. A memory you own that grounds every route. A real settlement rail that clears the work. Each layer is built, not borrowed.

Layer 01

A decentralized substrate

The intelligence underneath Rivier runs on Tenzro — a decentralized inference network with no single operator. Open models are served by many nodes, so the same model is reachable through several independent paths. If one node disappears, the next one answers; if a whole region goes dark, the routing layer picks a different one. The strongest available model is reached only when the work earns it.

Layer 02

Rivier One — our orchestration model

Rivier One is our own orchestration model. It decides which model handles each request, and answers itself when it can. It prefers the decentralized path for everything the network can answer, and reaches for a stronger model only when the prompt needs it — within your cost and privacy ceilings. Every route is auditable: tier chosen, latency, cost, any fallback.

Layer 03

The memory it routes against

Rivier holds a private, encrypted picture of who you are — your taste, your plans, the way you like things done — on Tenzro. Every route grounds in this. Open models on the network handle anything sensitive; a stronger model sees personalisation only, never the raw graph behind it, and never anything that identifies you.

Layer 04

Identity and settlement

One Rivier Identity binds who you are to how you pay. The accounting layer is one stable balance, seen in your own currency, with every confirmed action carrying an audit-grade record your finance team can reconcile. It settles for real on the same institutional rails the biggest banks and markets already use. Your balance isn't a number in an app. It's money that actually clears.

Specification

The system, plain.

• SubstrateTenzro decentralized inference network. Open models served by many nodes — no single operator can take the system down.
• MemoryEncrypted personal graph on the Tenzro network. User-owned, portable, never handed to the host assistant.
• ReasoningRivier One — our own orchestration model. Decides which model handles each request, answers itself when it can, and reaches for stronger models only when the work earns it. Cost, latency and privacy as ceilings per call.
• ActionScoped agent identity plus an institutional-grade settlement layer — book, buy and pay within user-set limits, on explicit approval.
• SettlementReal money that clears on institutional-grade settlement rails the biggest banks and markets already trust — not just a balance in an app.
• DistributionAn OAuth-protected MCP server. Same six verbs inside Claude, inside ChatGPT, and on the Rivier web. Connect once, used everywhere.
• ReceiptsEvery action carries an audit-grade record — what ran, what it cost, when, and who authorised it. The full detail is one click away whenever you want it.