Two years ago we started building something we had not seen anywhere else. A way to take the data your farm generates every season and use it to answer the questions that have always been answered by gut feeling: which spat performs best on your farm, under your conditions, in your water.
Today Spat Weather Analytics is Live!
As far as we know, the first tool of its kind to run this kind of live AI analysis directly on each farm’s own data. It reads your seedings, assessments, and harvests, cross references them with the weather your farm was exposed to during those periods, and surfaces the patterns that are actually there in your history:
- Which spat sources grew fastest.
- Which weather conditions lined up with better or worse outcomes.
- Which management decisions correlate with stronger performance.
- What your best farms are doing differently.
Available now for mussels, oysters, and seaweed. This is not a generic model built on industry averages. It is built entirely from your own farm records. The more data you have in the system, the sharper the analysis gets. Farmers who have been on Mussel App for six months or more will see meaningful patterns right away.
In this newsletter we’ll walk you through the dashboard and then we’ll show you a couple of important things to keep in mind to take the best advantage out of this feature.
Go to Stats > Spat Weather Analytics to get started.
Walking through the dashboard
1. Start with the summary
At the top you see headline numbers: total intervals, spat sources being compared, average daily growth, average harvest size, and weather coverage percentage. Below that, four cards give you the headline findings before you go into the detail: your top performing spat source, the strongest weather signal the app found, your best farm, and a data quality flag if coverage is too low to draw reliable conclusions.
2. See which spat has performed best
The best spat sources chart ranks your spat by performance score. Scroll down to the leaderboard for the full detail: raw score, adjusted score, average daily growth, peak wind and storm exposure, confidence rating, and an explanation per source. The adjusted score accounts for differences in farm, line, season, and weather so you are comparing spat on more equal terms.
A source with one seeding and a high score might have just gotten lucky. One with many seedings and high confidence is a much stronger signal.
3. Understand how weather has affected your farm
The weather driver insights section shows the patterns with the strongest association with your outcomes, written in plain language. For example: higher wave exposure in the first 14 days is associated with lower retention and yield.
Each insight tells you how many samples it is based on. Open any insight to see the individual growing intervals behind it. The weather driver correlations chart and table sit alongside it, showing the same information visually. Negative values mean higher weather readings tended to line up with lower outcomes. These are associations in your data, not proven causes, but they are worth knowing when planning a season.
4. Check growth across weather conditions
The weather buckets section groups your growing periods into bands by storm hours, peak wind, and peak wave, and shows the average daily growth in each band. A practical way to see how your stock has historically responded to different levels of weather exposure.
5. Use the forecast and watch list
The model flags seedings worth watching right now. Expected harvest date, amount, quality, and risk per seeding. Use it as an early warning list, not a harvest promise.
6. Look at your management decisions
Management factors takes the decisions you have recorded over time, density, spacing, submersion, stage path, floats, and looks at how they associate with performance scores. Each factor is broken down by band so you can see which ranges have tended to produce better results. These are associations, not instructions, but they can surface patterns that are hard to see when you are in the middle of a season.
7. Compare your farms and lines
Best farms and lines ranks your farms and lines by performance score and shows the conditions that coincided with the top results. The farm comparison table below gives you a side by side view of growth, yield, harvest size, temperature, and peak wind across all your farms.
8. Check the data behind the analysis
At the bottom you will find growth intervals, the raw growing spells that every score and chart is built from. This is the audit trail if you want to verify what data the system is working from. Environmental data sources shows which weather feeds are backing the report, currently Windy, providing modelled temperature, wind, waves, rainfall, and pressure.
A few things to keep in mind
A. This AI only works if your data is in the system
We cannot go back in time to collect weather data. The moment you join Mussel App we start collecting weather on your farm, and that is data nobody else has. If you want to see your own predictions in the future, the time to join is now. Farmers with six months or more of data will see meaningful patterns right away. The longer you have been recording, the more the analytics have to work with.
B. Data quality matters more than you think
With AI, what goes in is what comes out. We added a Data Quality Flags page at Stats > Data Quality that scans your records and surfaces anything that looks inconsistent or unusual, an assessment before a seeding date, a final harvest with no amount recorded, and so on. These errors are easy to miss but they affect what the AI learns and what it predicts. You will also see data quality flags directly on the dashboard when they are affecting your results. When you see one, it is worth fixing.
C. The AI learns when it gets things wrong
At the bottom of the Harvest Planner you will find the ML Harvest Log. Every time the AI predicted a harvest date and was off by two weeks or more, that case gets logged. We use those logs to look at why the prediction was wrong and teach the model to estimate better next time. It is not a list of failures. It is how the AI improves.
D. Stage reseeding
Also available in the Harvest Planner under Stage Reseeding. Shows which lines are ready to move to the next stage based on the analytics forecast, with a confidence rating and sample count per line. More on this in the next newsletter.
E. Explain Button
You can always click on the Explain button at the top right of each section. The system will tell you what that information is for or how it interacts with other sections.
Why we built this
Farmers already know a lot. The goal of this feature is not to tell you how to farm. It is to make the knowledge you have already built up, through years of seedings, harvests, and observations, easier to see, compare, and use when you are making decisions about the next season.
The data is yours.
This is just a better way to read it.
If you have any feedback, changes, updates that your farm needs let us know,
we’ll make it happen for you.
