I installed PostHog with the AI wizard to see your onboarding from the developer's chair.

The Pitch: Where I'd Start

I'd start with what I already built: a hands-on run of the PostHog setup wizard on a mock SaaS app, with the code public. See the full run →

The wizard is an AI agent, and that's the story to tell. It read my page components, found the signup handler and the call-to-action buttons, and wrote the tracking calls on its own. The whole gap between "I have a project" and "events are flowing" came down to one command and a browser login.

Developers don't believe a claim like that from a landing page. They believe it when they watch it happen on code that looks like theirs. So the launch I'd run is an uncut screen recording of the wizard on a real open-source repo, with the generated diff in the docs, and I'd judge it on how many installs reach a first insight. The wizard's newer self-driving mode is the bigger story after that, tested the same way first.

The Proof

The full run is on one page: the app, what the wizard did step by step, and the launch plan. Read it here →

I build the tools I market with. I run my own library of AI agent skills in Claude Code every day and publish part of it as open source, so an agent that writes code inside a developer's project is familiar ground.

I also write for a living. The MIA newsletter holds a 47% open rate on practical, show-the-work writing. At Reform, I repositioned the product and trial activation rose 34%, the same work a feature launch needs: say exactly what changed and who it's for.

Let's talk about PostHog.

If this is the right problem for you, I'd like to talk about it directly.