Python gives us a fast feedback loop: change some code, run it, see what happens. But once our applications depend on databases, queues, and other services, that feedback loop gets much slower. Code can work locally and still fail in production because of a configuration mistake or a mismatch between what the application expects and what’s deployed.
What if we could catch those integration problems before deploying?
In this talk, I’ll show how we’ve built an open source project that understands the infrastructure an application needs directly from its code. We’ll follow a change from local development to deployment, look at integration mistakes we can catch in seconds, and unpack how it works under the hood. You’ll leave with a different way to think about the boundary between application code and infrastructure, and what happens when developers and their AI coding agents can quickly get feedback on both, instead of waiting for deployment to find out what’s wrong.