Ship the Endpoint and the Dashboard

final deployment and the finished capstone

Part of: Ship & Monitor an LLM App (Capstone)

Every piece is built: a packaged request handler, a live HTTP route, structured logs, per-call cost, an aggregated dashboard, graceful error handling, and a budget guard. This lesson wires them into one running service, deploys it, and calls the capstone done. Finishing it lands the whole build in your Portfolio as a live, monitored LLM service you can point anyone to. Wiring it all together Each route is thin on purpose. It calls the functions built in earlier lessons instead of reimplementing anything. That's the payoff of packaging everything as reusable functions from lesson 1 on: the app is really three routes gluing existing pieces together. Deploying it Push the code to a small platform (Render, Fly.io, Railway) or a VPS with the Dockerfile from lesson 2, set ANTHROPIC API KEY as an environment variable (never in code), and start the process. Now you have a public URL. https://your-app.onrender.com/generate is a real endpoint anyone can curl, and /dashboard shows real usage. The done checklist Before you call it shipped, three questions, now with teeth: 1. Runs from a clean start : a fresh clone, one install command, one run command, no manual steps you forgot to write down.

Challenge: Final Health Check