AI discourse these days is all over the place. Depending on who you talk to, AI’s are absolute flash-in-the-pan junk, or they’re the best thing since sliced bread. I want to cut through the noise, though, and see for myself what someone can do out here on the bleeding edge. Thus, I’m setting myself a challenge: write a usable—and useful—application with Claude Code, from soup to nuts.

In a really broad sense, the history of observability tools over the past couple of decades have been about a pretty simple concept: how do we make terabytes of heterogeneous telemetry data comprehensible to human beings? New Relic did this for the Rails revolution, Datadog did it for the rise of AWS, and Honeycomb led the way for OpenTelemetry.

We recently had the privilege of hosting several industry experts and technology executives across platform strategy, SRE, and engineering enablement for breakfast at our Observability Day in London. We noted that they’re all facing the same fundamental tension: deliver faster, scale smarter, stay resilient, and somehow get ahead of what’s coming next. But how do you move fast without breaking things? And how do you prove the value of the things you don’t break?

I have a personal vendetta against “dashboards.” Not because they're not useful—I actually think they’re extremely useful—but rather because they're generally built with the wrong user in mind, then used by a completely different user, and for a different use case. 

You shipped your latest release. You tested it on emulators, QA devices, and the latest OS versions. But now it’s live and running on thousands or millions of real devices, across a jungle of screen sizes, hardware specs, OS versions, and network conditions. A user reports a crash on an old Samsung device over 3G. Someone else complains the app feels “sluggish” after updating. You dig through logs. Rebuild test cases. Ping the backend team. Try to reproduce. Yet, still no answers.

There are certain important metrics that every mobile app has in common. At Honeycomb, we have surveyed these metrics across iOS and Android, and have defined a set of Core Mobile Vitals we think every app developer should care about. The purpose of these vitals is similar to that of Core Web Vitals for web frontends.

I anticipated this would be a challenging time and that I would be exhausted. So, the plan became: do all the demanding things, take my sabbatical in May, and use April as an ‘in-between’ period with a bit less pressure. I would willingly step off the gas and let other SREs on the team cover pressing matters, as a sort of pre-game for my full month away.

There are many vendors, Honeycomb included, where actions on the application can emit a web request that goes to another service for coordination or tracking purposes. Many vendors have pre-built integrations, but some have a fallback that says “Custom Webhook” or similar. If you’re looking to create a full picture of your request flow, you would want these other services to show up in your trace waterfall.

I want to know what users are doing in my application. A distributed trace is the best way to show the data flow of one user interaction through my application, but it isn’t sufficient to show the overall user experience. 

Kubernetes is widely used for deploying, scaling, and managing systems and applications and is an industry standard for container orchestration. Google engineers originally developed Kubernetes as an open-source project. Its first release was in September 2014, and since then, it has matured into a graduate project maintained by the Cloud Native Computing Foundation (CNCF). With the complexities of scale and distributed systems, debugging in Kubernetes environments can be difficult. 

In this post, I’ll cover what synthetic monitoring is and show an example of how you can create a simple monitor using OpenTelemetry, .NET, and an Azure function. If you only want to see how it’s built, skip ahead to building a synthetic monitor.