The Data — Neo Genesis · 26 days, every reading logged

1.010 FG final gravity, stable

6.66% ABV the 666 mark

83.9% efficiency past target on day one

530 L batch the biggest collab yet

66 IBU 11 EBC colour

0.55 yeast k the yeast rewrote the model

0.000 hop creep across 3 days

ZERO diacetyl forced test, D19

**Fermentation — predicted vs actual**Gravity, pH, temperature and pressure across 22 days. The model said eight days to FG; WHC LAX got there in five.

**Daily gravity drop**Eleven points a day at peak. Hop creep is visible — and tiny — around D15.

**Attenuation & ABV**The climb to final gravity, apparent attenuation tracked against ABV.

**Temperature journey**Strike 71.3°C → mash steps → D-rest 20°C → dry-hop crash 12°C → cold crash 1°C.

**Brew-day gravity**First, second and third runnings through the mash, then pre- and post-boil.

**pH journey**5.27 at pitch down through fermentation; the dry-hop pH move that caught a mistake.

k = 0.55

The yeast rewrote the model.

The first fermentation model used the textbook constant k=0.213 and missed by 18 gravity points. WHC LAX ferments ~2.6× faster; refit to k=0.55 and the curve matched. Don't predict from fewer than three data points.

instrument layer

Six tools, eleven scripts, one publisher.

The AI built a calc bench (Brix↔SG, dilution, efficiency, FG-correction, Newton cooling), eleven plotting/analysis scripts, and an MQTT publisher feeding the tap-side display — 19 instruments for one batch. No invented numbers.

Raw data lives in the brew-day CSV and batches.json; the plots above are rendered straight from it. Numbers over branding.

Read the story →