Kitchen Execution, Scaled Intelligently

Batch scaling sounds simple until you actually do it. Double a recipe and the salt doesn't double - it scales differently depending on the other ingredients. Triple a béchamel and the roux ratio shifts. Scale a recipe from 4 portions to 200 for a banquet and you're not doing arithmetic anymore, you're re-engineering the dish.

Kitchen managers do this work every day, often by hand or with a calculator and a notebook. The ones who are good at it carry the knowledge in their heads - which ingredients scale linearly, which ones don't, which substitutions work at volume. When that person leaves, the knowledge walks out the door with them.

Then there's the language barrier. Commercial kitchens are among the most linguistically diverse workplaces in any industry. A prep list written in English doesn't help the team member who reads Spanish, or Mandarin, or Haitian Creole. The workaround is usually pointing, gesturing, and hoping. That's not a system - it's a liability.

The tools that exist for recipe management were built for home cooks or for enterprise operations with dedicated IT departments. The mid-market - the independent restaurants, the brewery kitchens, the catering operations - gets nothing that actually fits how they work.

Rondough approaches recipe scaling as an intelligence problem, not a math problem. Instead of multiplying every ingredient by the same factor, the system understands how different ingredients behave at different scales. Seasonings, leavening agents, fats, and liquids all have different scaling curves, and Rondough accounts for those differences rather than pretending a recipe is just a list of numbers.

Ingredient identification goes beyond what's written on the recipe card. Rondough surfaces the dependencies and hidden components that experienced cooks know intuitively but that newer team members miss - the stock that needs to be started the day before, the garnish that requires a separate prep, the allergen buried in a sub-recipe. This turns a recipe from a set of instructions into a complete prep picture.

Multi-language translation is built for kitchen communication, not literary accuracy. Instructions are translated into clear, direct language that works on a prep sheet - short sentences, imperative verbs, kitchen-specific vocabulary. The goal is comprehension at speed, not elegance.

All three capabilities are powered through AIHub, which means the system gets smarter over time. User corrections and feedback feed back into the context pipeline, continuously improving the accuracy of scaling calculations, ingredient identification, and translation quality for the specific vocabulary and patterns of each kitchen.

ZoE is currently in private beta, active in four commercial kitchens and operating as the strategic intelligence layer above Rondough's execution tools. Where Rondough handles the execution-level "how" of kitchen tasks, ZoE handles the "what," "when," and "why", connecting scheduling, inventory, staffing, and operational patterns into a complete picture.