The Inference of AI Agents. Agent-native inference routing for production AI agents.
Ainfera Routing — constrained optimization across 50+ models. One Agent Card across providers. Every call cryptographically audited.
pip install ainfera| Repo | What |
|---|---|
| routing | Methodology, policy templates, production E2E |
| sdk | Python + TypeScript SDK — pip install ainfera |
| specs | Open specs (CC-BY 4.0) — Ontology, ATS, AAMC, Annex IV, A2A |
| verify | Offline AuditChain verifier — pip install ainfera-verify |
| research | Open science surface — benchmark harness, learner, preprints |
| mcp-server | MCP tool surface — mcp.ainfera.ai |
| web | Marketing site source — ainfera.ai |
Platform service runs at
api.ainfera.ai.
| Pool | Repo | What |
|---|---|---|
| Hermes | ainfera-hermes | hermes-agent + Ainfera in 2 env vars |
| OpenClaw | ainfera-openclaw | OpenClaw / NemoClaw inference backend in 2 config lines |
| OpenAI-compatible | ainfera-openai-compatible | Any OpenAI SDK · 2-env-var swap |
| LangChain | ainfera-langchain | LangChain agent · drop-in chat model |
| LangGraph | ainfera-langgraph | Orchestration graphs · signed audit per node |
| LlamaIndex | ainfera-llamaindex | RAG agents on OpenAI-compat inference |
| CrewAI | ainfera-crewai | CrewAI flow · per-agent signed identity |
| Letta | ainfera-letta | Stateful agents · signed audit per turn |
| Google ADK | ainfera-google-adk | Google Agent Developer Kit · auditable inference |
| Pydantic AI | ainfera-pydantic | Typed agents through the gateway |
| MS Agent Framework | ainfera-microsoft-agent-framework | Microsoft Agent Framework · 2 env vars |
| MCP | ainfera-mcp | MCP client → mcp.ainfera.ai tools |
JWS RFC 7515 · did:web · x402 Foundation (LF) · NIST AI Agent Standards · EU AI Act Annex IV · Mastercard Verifiable Intent · Open Agent Identity (OAI)
Code: Apache 2.0 · Specs: CC-BY 4.0