AI & Technology
How AI Predicts When Your Sourdough Is Ready
Fermentation depends on dozens of variables. Here's how an AI model learns your kitchen to predict bulk and proof timing.
AI predicts sourdough readiness by learning the relationship between your kitchen's conditions (temperature, starter strength, hydration, flour) and how your dough actually ferments — then forecasting when bulk fermentation and proofing will be done. Instead of a fixed recipe clock, it gives you timing tuned to your specific bake.
Why fixed timings fail
A recipe that says "bulk for 4 hours" assumes one set of conditions. But fermentation speed depends on many variables at once:
| Variable | Effect on timing |
|---|---|
| Temperature | Biggest factor — doubles/halves speed |
| Starter strength | Faster if vigorous |
| Hydration | Wetter ferments faster |
| Flour type | Whole grain ferments faster |
| Starter percentage | More starter = faster |
| Salt | Slows fermentation |
A 4-hour bulk in a warm kitchen might need 8 hours in a cold one. Fixed times can't account for that — which is why beginners over- and under-proof.
How the prediction works
- You log the bake — flour, hydration, starter amount, and kitchen temperature.
- The model estimates a fermentation curve for those conditions.
- It refines over time. Each bake you log teaches it how your starter and kitchen behave, so predictions get more accurate.
- It outputs a window, not a single number — "bulk likely done between 5:30 and 6:15 PM," with cues (rise %, jiggle, poke test) to confirm.
What it can and can't do
| Can | Can't |
|---|---|
| Estimate timing from conditions | Replace your judgment entirely |
| Learn your kitchen's patterns | Bake the loaf for you |
| Flag likely over/underproofing | Know what it hasn't been told |
The best results come from combining the AI's prediction with the physical signs — the model points you to the right window; you confirm with your eyes and hands.
Frequently asked questions
Does it actually know my kitchen?
It learns from the data you log — the more bakes, the better it models your specific starter and conditions.
Can it predict proofing too?
Yes — the same approach applies to the final proof, accounting for temperature and how far bulk progressed.
Is it a replacement for learning?
No — it's a guide that accelerates learning by showing you how conditions change timing.
This is exactly what SourdoughAI does — it learns your kitchen from your logged bakes and predicts when bulk and proof are actually done, so you stop guessing.