AI agent for chefs
Recipe Scaling Test and Revision Agent
Produce a scaled recipe that tastes and behaves like the original at the new size.
What it does
A sauce that works in a 2-quart pot is salty and thin when made for 400 guests, because cooking times, evaporation and seasoning do not scale in a straight line. The agent scales the recipe to the target yield and flags steps that rarely scale evenly, such as salt, spices, cooking time, reduction, leavening and oven loading. It plans a test batch, either full size or a mid-size trial, with clear things to record: time, temperature, taste, texture and yield. After the cook records the results, the agent compares them with the target and proposes a revision, for example less salt or a longer simmer. It updates the recipe and plans another test if the result is still off. It records the final scaled version with notes. The chef approves the revised recipe. Edge case: a baked item where the pan size changes.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- A recipe needs a new batch size
- Read the recipe, target yield and equipment sizes
- Scale quantities and times, flag steps that do not scale evenly and adjust them by rule
- Plan a test batch and a recording sheet
- Chef approves the test planThe agent waits here for your OK.
- Cook records the test results
- Do taste, texture, time and yield match the original?If not: propose revisions from the results and plan another test. Back to step 3.
- Write the final scaled recipe with notes and cost
- Is the cost per portion within the recipe's target?If not: adjust ingredients or portion size and test again. Back to step 3.
- Chef approves the revised recipeThe agent waits here for your OK.
- Scaled recipe stored
How it decides
Linear scaling is the start. Steps known to scale unevenly are adjusted by rule, and each test result is compared with the original's taste, texture and yield.
- Scale salt and spice to 75 percent of linear for batches over 4 times
- Lengthen cooking times by oven load, not by a straight ratio
- Run at most 3 test rounds before asking the chef
- Record yield by weight
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Scaling rules for spices and salt
- Number of test rounds (default 3)
- Target description
- Cost target
- Recording sheet format
What keeps you in control
It always asks you first
- Chef approves the test plan
- Chef approves the final recipe
Hard limits
- Never change allergens without the chef
- Record every test for traceability
It stops when
- Done: scaled recipe approved
- Stop: three tests fail, return the results to the chef
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide