These examples walk through real-world use cases end to end. Each one starts from a common customer goal and takes you through the specific steps to achieve it.

Connect your first AI provider

Add OpenAI, Anthropic, or any supported provider. Create a logical model and API key. Make your first call through the gateway in under 5 minutes.

RAG pipeline with knowledge base

Create a knowledge base, upload your documents, connect a workspace agent, and query it via chat. Ground your AI answers in your own content.

Canary routing between providers

Split traffic between two model providers using configurable weights. Safely migrate from one model to another without a big-bang cutover.

Deploy an open-weight model

Host Llama, Mistral, or another open-weight model on your own GPUs. Register it as a provider in the gateway and call it through the standard OpenAI API.

​
Where to start

If you’re new to ManyLayers, begin with Connect your first AI provider — it covers the core concepts (providers, logical models, teams, keys) that all other examples build on. If you already have providers connected and want to build a RAG application, jump to RAG pipeline with knowledge base. If you’re evaluating ManyLayers for multi-provider resilience, see Canary routing between providers.