Nirmos Documentation
Build, operate, and scale AI applications with the Nirmos platform.
Nirmos provides one infrastructure layer for model access, routing, caching, agents, workflows, observability, vector retrieval, and authentication.
Build with the TypeScript SDK
Use the official SDK for normalized Gateway requests, streaming, embeddings, images, and managed prompts.
Nirmos TypeScript SDK
Explore the stable application interface.
SDK quickstart
Send a chat request and use a managed prompt.
Start with the AI Gateway
Connect an application through one OpenAI-compatible endpoint, then add provider and routing policy without changing the application contract.
Connect your first application
Send a model request through Nirmos.
Configure provider fallback
Recover from provider errors and rate limits.
Read the engineering blog
Production AI infrastructure notes from Nirmos.
Platform areas
| Area | Purpose |
|---|---|
| AI Gateway | One request interface across providers |
| Routing | Select models by policy, health, latency, and cost |
| Caching | Reuse exact or semantically equivalent responses |
| Observability | Trace attempts, tokens, latency, failures, and spend |
| Agents and workflows | Run durable tool and model execution paths |