Running a recipe site means writing, photo work, ingredient lists, scaling, shopping lists, and a hundred small chores. The cooking is the fun part. Most of the rest is busywork. That is where AI earns its keep.

What AI does for this blog

I draft a recipe in a notebook the way I actually cooked it: rough quantities, steps in the order I did them, notes in the margin. AI helps me turn that mess into structured data: clean ingredient entries with units, numbered steps, aisle assignments for the shopping list, and timer values where they matter. The recipe itself is mine. The boring transcription is not.

It also helps with the dev side. The ingredient scaler, the cook-along mode, the aggregated shopping list across recipes — all of those features were planned and refined with an AI pair. I still review every change, run it locally, and decide what ships. AI is the keyboard, not the chef.

What AI does not do

It does not invent recipes for me. It does not taste anything. It does not decide that a dish needs more lemon. The voice on this site, the opinions about which plate to use and when, the photos of food I actually ate — that all comes from me. AI just keeps the plumbing tidy so the cooking can stay the point.

If you are running a small project of your own, I recommend the same split. Hand the boring stuff to the model, keep the judgment for yourself, and spend the saved hours on the part you love.