Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Gatewayz: Multi-LLM Routing and Cost
This talk details building Gatewayz, a system for routing traffic across multiple LLM providers, tracking cost, and handling retries, all generated with Claude Code.
Gatewayz started as a way to solve a personal problem: I needed a clean, fast, and observable way to route traffic across multiple LLM providers without rewriting my stack every time a new model dropped. This talk walks through how Gatewayz was built from the ground up, completely with Claude Code over the course of a 2 week sprint.
I’ll demo the architecture, how requests are normalized and routed, how latency and cost are tracked in real time, and how I avoided turning the system into an un-debuggable distributed nightmare.