Getting Started with agentgateway Series 01: Deploying and Configuring Your First LLM Request
Overview
Go from zero to a working LLM gateway in about five minutes. In this first video of the agentgateway series, we spin up a Kind cluster, install agentgateway, configure OpenAI as our first provider, and send a chat completion through the gateway — all running locally on Kubernetes.
What We Cover
- Installing a Kind cluster for local Kubernetes
- Deploying Kubernetes Gateway API CRDs + agentgateway CRDs
- Installing the agentgateway controller (v1.0) via Helm
- Creating a Gateway proxy to handle traffic
- Configuring OpenAI as an LLM backend with secure secret storage
- Setting up an HTTPRoute to map requests to the backend
- Sending your first request — the model is auto-filled by the backend, so clients don’t need to know which model they’re hitting
Key Takeaway
One of the nicest patterns for platform teams: the backend already has the model configured, so agentgateway fills it in automatically. Your clients don’t even need to know which model they’re hitting.
What’s Next
This is just the beginning. In the next videos we’ll set up more providers (Mistral, Anthropic, self-hosted models), then get into failover, rate limiting, and cost tracking.