An Agent Skill for Swift Testing · Using Gemini in Xcode · Automating Codex & The Agentic Loop

While enjoying time off with the family last week during a ski holiday, I found myself re-evaluating the status quo. It's always good to take a step back and revisit what you're doing.

I figured and asked myself:

If you think about agents like colleagues, you would expect them to continue working while you're off. It's the same for weekends, basically any moment you're not writing code yourself.

So I started reading up on automations. Codex offers automations, we can use tools like Zapier, or go all in on OpenClaw (now part of OpenAI, congrats Peter!).

Either path starts by connecting the dots, giving AI eyes. My current tech stack involves RevenueCat for purchases, Amplitude, Plausible, and App Store Connect for other tracking measurements. Agents can already access my code, so what if they can connect the dots?

I did not yet automate anything, but I did start a new paywall experiment based on code analysis combined with insights from RevenueCat and Amplitude. Simply because I can now simply ask:

This all started by me evaluating where I spent my time. I love looking at my charts, but it's time consuming and I often don't have as much product knowledge as my agents. This truly feels like a next step in my workflow and the future will be about automating this monitoring and running new experiments autonomously.