OohBae Kitchen Advisory Board
The Setup
Running a cottage bakery means answering questions that Google can’t. How many batches of ube crinkle cookies do I need for a 50-person event? If butter goes up $2/lb, what happens to my margins? Can I add brownies to the menu without doubling my bake day? What’s the shelf life if I switch packaging?
These aren’t strategy questions. They’re kitchen math — specific to our recipes, our costs, our production schedule, and Wisconsin’s cottage food laws. I needed a tool that knew all of that and could do the math faster than I can.
The Build
I built a second Claude Cowork skill — a 26-expert food operations advisory board. This one is specialized for the bakery. It knows every recipe (ube crinkle cookies, brownies, muffins), every batch multiplier, every ingredient cost, and every production constraint.
The experts include a recipe scaling specialist, a food cost analyst, a packaging advisor, a cottage food law expert, a shelf-life consultant, and a production scheduler — among others. Each one has access to our actual product data, pricing, and cost breakdowns loaded from our brand bible files.
When I ask “Can we take a 100-cookie catering order for next Saturday?” the board doesn’t just say yes or no. It tells me how many batches, what ingredients I need to buy, what it’ll cost to produce, what to charge, and whether it fits our production schedule.
The Mess
The first version couldn’t do math reliably. AI is great at reasoning about business strategy but surprisingly bad at multiplying ingredient quantities across batch sizes. I had to structure the skill so it referenced actual recipe data (grams of ube extract per batch, cost per pound of butter) rather than trying to estimate from memory.
Getting the cottage food law expert right was also tricky. Wisconsin cottage food laws have specific rules about what you can sell, where you can sell it, and how much. The skill needed to know those constraints cold — not approximate them.
The Result
The Kitchen Board handles questions that used to take me 20 minutes of spreadsheet work. Batch planning for events, ingredient cost projections when prices change, new product feasibility. Miles and Zelda are learning the business side by watching me ask these questions and seeing the answers broken down.
It’s also a teaching artifact. When I show people what Claude Cowork skills can do, the Kitchen Board is the example that makes food business owners lean forward. It’s not abstract AI strategy. It’s “tell me how much butter I need for Saturday.”
The Takeaway
The best AI tools aren’t the impressive ones. They’re the ones that replace the spreadsheet you were too tired to update.
How It’s Built
- Platform: Claude Cowork (skill architecture)
- Context: Loads recipes, costs, production data from Ube MKE brand bible
- Experts: 26 defined personas covering food ops, law, costing, production, packaging
- Data: Real ingredient costs, batch multipliers, pricing tiers
- Cost: Included in Claude subscription