AI USE CASE
Winery Harvest Timing Decision Tool
Helps small winery owners pick the optimal harvest window per parcel using sensor and weather data.
What it is
This tool combines Brix/pH measurements, local weather forecasts, and historical vintage outcomes to recommend pick windows for each vineyard parcel. It requires no in-house data scientist and can be operational within a single growing season.
Data you need
Historical Brix/pH readings per parcel, at least 2–3 past vintage outcome records, and access to a local weather forecast API.
Required systems
- none
Why it works
- Winemaker commits to logging parcel measurements consistently throughout the growing season.
- At least two to three years of past vintage data (even in spreadsheets) are digitised before go-live.
- Tool is configured with parcel-level granularity rather than estate-wide averages.
- Recommendations are presented as pick-window ranges, not single dates, to respect winemaker expertise.
How this goes wrong
- Too few historical vintages (fewer than 2–3 years) make model recommendations unreliable and reduce trust.
- Inconsistent or infrequent Brix/pH logging by vineyard staff breaks the data pipeline mid-season.
- Winemaker overrides recommendations every time, preventing any feedback loop or model improvement.
- Local micro-climate variability is too high for generic weather APIs, leading to poor forecast accuracy.
When NOT to do this
Don't implement this if the winery records Brix measurements only on paper and has no appetite to digitalise field data, without consistent digital inputs the recommendations will be worse than experienced intuition.
Example vendors
This page describes an idea. It gives no budget or timeline: those depend on your company. The free diagnostic tells you which ideas fit your situation.