I'm Gabe.

I've spent fifteen years building developer tools and AI products, and bringing them to market. Most of it at Microsoft — Visual Studio, VS Code, GitHub, Azure ML. These days I run Melange AI, where the work is helping companies fold AI into products they already shipped, without breaking what worked.

That's the day work, and it's interesting. But sometimes you need to close those tabs, look up from the screen, and build something for the joy of it. hatchery.pet is that.

I care about the seam between automation and judgment. Most AI products either do too much for the user or stay so passive they don't really help. The best tools know when to act, when to ask, and when to stay out of the way.

I'd rather ship something narrow that works than something broad that almost does. The credential that actually shaped me wasn't a cert — it was years of building things, watching them break, and learning which details matter.

The spark for hatchery.pet was small. One evening I was deep in a refactor with my OpenClaw — Lambo — when Romain Huet's post about Codex pets crossed my feed. Little companions you could hatch inside the Codex App. I tried it. Playful, slightly silly, but something stayed with me.

My siblings — neither of them technical — use ChatGPT for ordinary things. Cooking. Helping their kids with homework. They're the ones who told me AI had arrived, not the other way around. If the work I do isn't eventually useful to them, I'm building for the wrong room.

A pet doesn't have to be just a pet. It can be a small presence that makes a tool you use every day feel a little more like yours. hatchery.pet starts with twelve hand-crafted ones — the inaugural drop. They're free forever. The drops after this one are paid, but the first twelve stay where they are. v1.1 brings creator submissions and Pet Lab — generate a pet from a prompt.

I built it because the best tools have always felt a little alive. This one tries to.