From one aronia cook to an Ingredients Studio¶
A few weeks back I put a few ingredients together and came up with an aronia Maillard preparation. Very soon after, prompted by brewers more experienced than me, I realised I did not really understand the chemistry, so I decided to research it in more depth.
That is where the idea of a skill to plan a Maillard cook first appeared, and it has been growing ever since. I have refactored the workflow many times already. The skill became two, which did not quite work: they overlapped, and each pulled and fed ideas in its own direction. The orchestrator came next, and it has kept evolving into what feels like a fairly complete ingredient analysis suite, the Ingredients Studio.
How it fits together¶
A router takes each request. In explore mode it runs a preflight, pauses, catalogues and triages, then gives me an overview to choose from; in directed mode it goes straight to the in-depth tasks. Specialists do the detailed work, shared records (registry, register, gaps backlog) keep track across runs, and the results land as designs and analysis pages.
Along the way came odd calculators too, such as one to assess cyanide levels when adding fruit pits to beer.
Where it stands¶
I have used the Ingredients Studio, but I have not run enough cooks and ingredient analyses yet to know whether the information is of little or great use. Next up is a bland(ish) beer I made with some grapes, and I want to try the Aged Wine Depth cook on it. The Maillard cooks are one of the most interesting things to come out of this home experiment with AI.
Previous: Writing it all down, with help · Next: The Cultures Studio, and using the tools in anger

