Reference · CC-BY 4.0
Ainfera Ontology v1.0.
The canonical vocabulary behind every Ainfera surface — the layers, entities, agents, and stakeholder roles that name the product. One vocabulary, used everywhere: the API, the dashboard, every audit log.
Five horizontal layers, one stack.
Each builds on the one below it.
| Layer | Name | Summary |
|---|---|---|
| L1 | Identity | Verifiable Agent identity. Agent Cards, signing keys, fingerprints. |
| L2 | Routing | One API call resolved to the right Provider and Model under a Tenant's policy. |
| L3 | Settlement | Wallet balances, per-Inference debits, signed Receipts, spend policies. |
| L4 | Audit | Hash-chained AuditEvents an Auditor can verify offline. |
| L5 | Governance | Regulatory event recording and Annex IV-compatible technical documentation. |
Core entity catalog.
Each entity has one canonical definition.
- Tenant
- The account that owns Agents, Wallets, and Audit Chains. The billing and ownership boundary.
- Agent /concepts/agent →
- An autonomous software actor that runs Inferences.
- Agent Card /concepts/agent-card →
- A signed identity credential minted for an Agent.
- Inference /concepts/inference →
- A single model call — signed, settled, and recorded as AuditEvents.
- Provider
- An upstream inference vendor Ainfera routes to.
- Model
- A specific inference model offered by a Provider.
- Wallet /concepts/wallet →
- A Tenant's balance; Inferences debit it.
- Receipt
- A signed record of a Wallet debit for a specific Inference.
- Spend Policy
- A Tenant-defined rule set on an Agent: caps, allowed Providers, allowed Models.
- AuditEvent
- A single signed record in an Audit Chain.
- Audit Chain /concepts/audit-chain →
- The hash-chained sequence of AuditEvents for an Agent.
Public-facing agents.
The agents Ainfera runs in production.
Manwe
Founder agent
The founder's personal AI brain, running in production on Ainfera. Not an Ainfera product — the infrastructure underneath.
Varda
AI co-founder
Ainfera's AI co-founder. Drafts, plans, and reviews — every Inference signed and audited like any Tenant.
Yavanna
Community agent
Ainfera's customer and community agent. Public-facing; its live audit feed comes online D14+.
Ainfera operates additional internal agents (Namo, Aule, Tulkas) that are not public-facing.
Stakeholder roles.
- Tenant
- Owns Agents, Wallets, and Audit Chains. Sets spend policies. Bears billing.
- Agent
- Acts on behalf of a Tenant. Holds an Agent Card. Runs Inferences.
- Provider
- Serves Inferences upstream. Named on every Inference it fulfills.
- Auditor
- Reads and verifies an Agent's Audit Chain. Can replay the hash chain without trusting Ainfera.
How entities connect.
Compose, don't invent.
The Ontology is designed to interoperate with these specifications.
- ATS v1.0 — Agent Trust Spec →
Ainfera's open specification for verifiable Agent identity.
- AAMC v1.0 — Agent-to-Agent Model Context →
Advisory for safe cross-Tenant context exchange.
- EU AI Act / Annex IV →
How Ainfera output maps to Article 12 and Annex IV documentation.
- NIST AI Risk Management Framework ↗
Risk-management vocabulary Ainfera's audit surfaces align to.
- Mastercard Agent Pay ↗
Agent-payment patterns informing the settlement layer.
- OpenAI Agents Spec ↗
Interoperability reference for Agent definitions.
- x402 Foundation ↗
Payment-required protocol patterns for machine settlement.
Glossary preview.
A preview of the A–Z reference. See the full Glossary →
- AAMC
- Adjudication and Auto-Modification Council — Ainfera's v1.1 deliberation system. 5 frontier Voters (Claude Opus 4.7, GPT-5.5, Gemini 3.1 Pro, Grok 4, Mistral Large 3) × 5 Lenses (Safety, Reliability, Quality, Performance, Reputation). Quorums: 3-of-5 advisory, 4-of-5 binding, 5-of-5 + Council for self-modification.
- Agent
- An autonomous software actor that runs Inferences on Ainfera. Every Agent holds an Agent Card and acts under a Tenant.
- Agent Card
- A signed identity credential minted for an Agent. Proves that an Inference request genuinely originates from a known Agent.
- Annex IV
- The EU AI Act technical-documentation annex. Ainfera emits Annex IV-compatible documentation so customers building regulated systems inherit it.
- ATS
- Agent Trust Spec v1.0 — Ainfera's open specification for verifiable Agent identity and signed Inference.