RoastRecipeAdvisor – Turning Your Roast History into Actionable Learning Loops
RoastRecipeAdvisor is a beta web app for Aillio Bullet users who roast from RoasTime recipes. It helps turn roast history into concrete recipe-learning loops: compare curves, understand what changed, preview recipe corrections, and test small step changes before committing them.
If you roast on the Aillio Bullet using RoasTime recipes, you’ve probably spent time scrolling through old roasts wondering what you actually changed last time — and whether it helped. Instead, the Roast Dashboard shows which roasts are on track, which have recommendations, which are ready for follow-up, and which need attention next. The app tracks the loop after a recommendation: whether the recipe was updated, whether the next same-bean roast is a clean follow-up, and whether the result improved.
Roast Dashboard
Overlay mode lets you compare the current roast against prior same-bean roasts, so you can see whether a curve problem is a one-off, a setup issue, or a repeatable recipe pattern.
Overlay Mode
Instead of simply pointing out problems, RoastRecipeAdvisor identifies the most likely root cause and recommends the next recipe change to test. Whole-curve analysis looks for the main problem first: setup/preheat, FC-transition dip, repeatable same-bean pattern, endpoint issue, or equipment/sensor conflict.
Whole-curve analysis
The goal is to recommend the next concrete roast test, not bury you in vague observations.
Whole-curve recommendations
When the app recommends a recipe change, you can view that recommendation directly on the graph and see where the change is intended to affect the curve. Not just “move P8 from X to Y,” but “here’s why, here’s where, and here’s the expected curve effect.”
Recommendations on the graph
Micro-adjustment mode lets you experiment with moving, adding, or removing recipe steps and preview the expected effect before changing your RoasTime recipe. Click on a power or fan transition and drag it to see the effect on the graph. This turns recipe tweaking into a logged learning loop instead of guesswork. It remembers your experiment and carries the test forward to your next roast.
Micro-adjustment drag preview
Finally, the app lets you Export for AI review. It can package the roast, curve data, recipe actions, markers, and context into a structured export you can paste into an AI tool for a second opinion.
Export for AI review
I’m looking for beta testers who roast on the Bullet using RoasTime recipes and are willing to give feedback on whether the recommendations are clear, useful, and actionable. If you already use recipe-driven roasting and want a better way to learn from each roast, I’d love to have you try it.
Try it here → RoastRecipeAdvisor
Happy roasting,
Carl