§01 · The brief, in one sentence
I had not picked a coffee cherry before the morning of the first harvest. The crew did the picking. I did the math, the freight, the 4 a.m. call about a generator. This is the field journal of week one: six days, eight pickers, two equipment failures, and the things we wrote down so the second harvest is less interesting than this one.
§02 · The first picking week
The crew started on a Tuesday at 6:14 a.m., twenty-six minutes after first light, with eight pickers on retainer from the year-zero stipend. They worked the upper boundary first, where ripening was a week ahead of the lower rows. By Friday we had moved through Block A. Saturday we paused for rain, which the crew predicted before the weather radio did. Sunday and Monday we finished the lower blocks and the outliers in Block C.
Selective picking (only ripe, only red, only the cherries that detach with a quarter twist) is what makes a specialty cup. The crew picked at roughly forty kilograms per person per day, well below the regional commodity average of seventy. The difference is in the basket: ours are 96 to 98 percent ripe by visual sort. Commodity baskets often run in the 70s. The premium per kilogram lands in the cup; the slower pace lands in the wage.
Forty kilograms a day is the wage of selective picking. The cup is what pays it back.From the field notes, harvest week 1
§03 · The wet-mill, and what we got wrong
The wet-mill is the heart of a Kenyan double-wash protocol: depulper, fermentation tanks, washing channels, soak tanks, then out to the drying beds. We built ours modular and small: 600 kg/hour depulper, three cement fermentation tanks at five hundred liters each, a forty-meter washing channel, two soak tanks. Sized for ten acres at full yield, not aspirational scale.
We got three things wrong. They're written below in the order we noticed them.
We sized the backup generator off the published wattage of the depulper alone. We hadn't accounted for the floodlights at the drying-bed shed, the water pump at the upper boundary, or, more obviously, the fact that two of those things ran simultaneously during a rain event. Day three, the generator browned out at four in the morning. We bought a second one, then a properly sized first one. The original is now the redundancy.
The depulper had a 600 kg/hour rating. The 600 was true at temperatures below 24°C, with cherry harvested in the last 12 hours, with the manufacturer's recommended pulp clearance. We were running it at 28°C, with cherry harvested 16 hours ago, and we hadn't adjusted clearance. It jammed twice on day one and chipped a part on day three. New rule: stress-test the depulper at production volume before harvest, not during.
The fermentation tanks ran for the entire first week without an in-tank pH meter. We were measuring pH by paper strip, which gives you 0.5 pH resolution at best, useless for fermentation calls. The proper meter sat in customs for fourteen days. We held the first lots a day longer than ideal because we couldn't measure them precisely. The cup was fine, but the protocol was a guess. Order the meter first next year, with two weeks of customs slack on top.
§04 · Drying beds: the hand-turn call
The choice between hand-turning and mechanical drying is not subtle. Mechanical is faster, more uniform, and capital-intensive. Hand-turning is slower, less uniform, and labor-intensive. Both produce a coffee that meets specialty grade. They do not produce the same cup.
We hand-turn. The crew rakes the beds three times a day in the first week, twice in the second, once in the third. Every parchment bean rotates through every part of the bed: edges (cooler, faster), middle (warmer, slower), top of pile (sun-exposed), bottom (shaded). The variance produces a complexity in the cup that mechanical uniformity flattens. It is also harder to argue about and easier to mess up. See the sister piece Why we still hand-turn the drying beds, when it's published.
§05 · What ships, what we'd do differently
The first harvest produced enough green for a single export container: twenty-two 60kg bags of the high-elevation peaberry sort, plus the larger main-lot tonnage that moves through the cooperative. Spec sheets, processing logs, and cupping forms ship as a single PDF dossier with every sample.
The list of things we'd do differently next year is short and not embarrassing: size the generator first, stress-test the depulper at production volume in August, order long-lead lab equipment with two weeks of customs slack, and start the picker retainer three months earlier so the year-one rhythm is the year-two rhythm. None of it is glamorous. All of it is the difference between an interesting first harvest and a calmer second one.