Lab Notes · Development timeline
I started working with burnt honey in 2022, before I made this ice cream. A chef colleague later shared a custard base with me, and I kept changing it until it became the formula I use now. Looking back, the larger change was in how I thought about flavor, structure, and repeatability.
-
I started with burnt honey, not ice cream.
My first notes focused on burnt-honey butter. My attention was on the honey before frozen dessert entered the work. I was learning how far I could cook it and how sweetness picked up roasted, bitter edges. Later use in sauce and ice cream turned burnt honey into a working technique for me.
-
I began with a custard base from a chef colleague.
The base used milk, cream, sugar, salt, and four yolks, with the eggs tempered much like crème brûlée. Converting the cups and yolk count into grams changed my relationship with the recipe. Cups let me reproduce the custard; weights let me separate its parts. I could adjust one component, watch the result, and build my own testing process.
-
Versions 1 through 5 moved me away from pure custard.
I lowered the yolk and cream, then introduced dextrose, nonfat milk powder, and dedicated stabilizer and emulsifier ingredients. I was beginning to treat sweetness, dairy solids, fat, egg, and texture as different controls. Some versions became too thick and grainy. Version 5 improved after I reduced some of the structure. I learned complexity and control were not the same thing.
-
Versions 6 and 7 defined my burnt-honey limit.
Version 6 had the texture and creaminess I wanted, although the dairy flavor initially seemed too far forward. The balance tasted better after it set. Version 7 replaced the dextrose with more burnt honey, but the ice cream became thicker and the honey flavor went too far. A chef colleague also preferred Version 6. Together, they taught me to judge the finished set, not just the draw, and to recognize when the signature flavor had begun working against the formula.
-
Version 8 turned the work into a production routine.
I moved back toward a more yolk-forward formula and started tracking the process as carefully as the ingredients. I portioned the base into bags, controlled the water-bath step, watched the churn, tracked the draw temperature, cured the ice cream, and ran a melt test. I began to understand the formula as a sequence rather than an ingredient list. Portion size, heating, chilling, churning, extraction, and curing all shaped what reached the table.
-
2024–2025
I stepped away from the recipe.
After Version 8, I stopped working on the ice cream. I made no new versions in 2024 or 2025. When I picked it up again in 2026, I had some distance from the custard experiments that shaped the earlier versions and a clearer idea of what I wanted the ice cream to become.
-
I rebuilt the ice cream as a hybrid formula.
I kept egg yolk, but I gave other ingredients more specific jobs. Burnt honey, toasted cream, and toasted nonfat milk powder formed three browned parts of the flavor. Dextrose, sucrose, milk solids, and a small stabilizer-emulsifier system gave me separate controls for sweetness, texture, and service. The custard ancestry is still visible, but the structure no longer depends on yolk alone. Reducing the formula to a 100-gram control also gave me one clear base for scaling.
-
July–August 2026
I scaled it, photographed it, and wrote the process down.
I worked through a batch and recorded cream, honey, milk powder, assembly, portioning, and churn. Writing the process forced me to name choices carried through repetition and kitchen habit. Looking across the versions showed me how uneven development can be. The ice cream carries the original custard idea, restraint from failed versions, and service habits.
