Tracking Yield Potential · Beta

The Yield module transforms your field observations, your field counts inspired by AWRI methods, plot history and local weather into progressive yield estimates, plot by plot. Each estimate is accompanied by a confidence level and a line-by-line explanation of the factors that influence it.

ŒnoT makes field data speak: your dictations, photos, counts, weather observations and harvest histories become usable signals to monitor harvest potential, plot by plot.

Status: private beta — developed in iteration with a consulting oenologist and a winemaker in the field. Access on request from page d'accueil.

A transparent agronomic model, not a black box

The objective is not to replace the professional eye, but to make the estimate more structured, more traceable and easier to justify. For each plot, ŒnoT combines:

1. Baseline grape variety: reference yield for the grape variety planted
2. Appellation objective: ceiling of the specifications (INAO) or objective that you set
3. Planting density: number of plants per hectare
4. Age of the vines: coefficient depending on the maturity of the orchard
5. Observed signals: 3-level extraction on your voice dictations (confirmed observation / contextual mention / no usable signal) to limit false positives on mildew, powdery mildew, botrytis
6. Plot calibration: exponential smoothing learned from your past harvests — each harvest entered refines the estimate for the following year

The plot sheet displays the details of each factor and the associated confidence level. You can justify the final figure line by line to a client.

Assisted counting inspired by AWRI methods

The AWRI (Australian Wine Research Institute) method consists of counting clusters in several areas of a plot. ŒnoT guides you via an AI chat:

• The chat asks you zone by zone: number of vines counted, number of clusters
• Automatic calculation of extrapolation to the plot (average clusters/vines)
• Auto-pause if you stop mid-session, resume at the stop point
• Minimum 3 zones for an estimate to be considered “Field” (reliability ±10%)
• Strict mode: if you mention mildew, weather or other observations, the AI politely reframes and sends these notes back to the Observations module

Starting a count: instructions for use

From the mobile app (in the field)

1. Open the My Domains tab, then select the domain concerned
2. In the list of parcels, locate the parcel to count. The Counting button is displayed when swiping to the right on the plot map (or via the scale icon at the bottom of the map)
3. Tap Counting — the “Counting · [plot name]” screen opens with an AI chat and a green “Yield counting mode” banner
4. The chat starts the conversation. For each zone, type your statement in natural language, for example: 10 ceps, 132 grappes
5. The cat saves the zone (“✓ Saved zone: Zone 1 · 13.2 clusters/vine”) and asks you for the next one. The top banner indicates the number of zones recorded in real time
6. After 3 minimum areas, the chat suggests finalizing. Answer “finish” (or similar)
7. A Diff card is displayed with the comparison before/after counting (yield range, reliability, tier). Press View file to return to the details of the plot, already updatedIf you interrupt the session, ŒnoT offers to resume it from the stopping point the next time you open it.

From the web interface (at the office)

1. Log in to web.oenot.io and open the Yields tab in the left menu
2. The Domain view is displayed by default: KPIs at the top, “Plots to monitor” table sorted by priority (red alerts = urgent, orange = vigilance, green = normal situation)
3. Click on a row in the table to open the Parcel detail (or use the Parcel dropdown selector on the Detail tab)
4. In the right column, locate the Field Counting card. Click on the green button Start a count: an inline chat opens in the card
5. Enter your statements zone by zone in the chat text box (same format as on mobile: 10 ceps, 132 grappes). The Send button validates each message
6. After 3 zones and the finalize command, a Diff card appears in the chat (before/after with directional arrow). The plot sheet is automatically refreshed (Estimate map, Evolution over the season graph, History of counts table highlighted green)

Automatic weather (sensorless)

From the GPS of the last observation, ŒnoT automatically recovers the useful indicators:

• Cumulative rain, degree day base 10, Huglin index, hot days and dry days (Open-Meteo archive, D-30)
• Live weather + 5-day forecast (OpenWeather, 10 min cache)
• Automatic update via weather pre-cache cron

AI vision on photos of clusters

The photos of clusters added to your observations are analyzed by GPT-4o-mini to support field counting (load, health status). This analysis does not replace counting, it contextualizes it.

The two web views

Domain view (default)

• Multi-plot panorama with aggregated KPIs (monitored plots, surface area, domain estimation, alerts)
• “Plots to monitor — by priority” table
• Export CSV button at the top of the table → downloads a oenot-rendements-domaine-YYYY-MM-DD.csv file

Parcel detail

• Estimate map: range in hl/ha + confidence level
• Graph Evolution over the season
• Table History of counts
• Card Field counting (AWRI cat) and card Actual yield (calibration) in right column

Conventional and organic recommendations

Actionable recommendations are available in conventional and organic options, with sources and usage limits displayed next to each recommendation. It is up to you to check the regulatory compliance applicable to your context before any application.

Actionable Pathogenic Factors

When a pathogenic factor (mildew, powdery mildew, botrytis, etc.) weighs on the potential of a plot, ŒnoT makes it actionable: you can create a reminder directly from the identified factor, to plan the intervention at the right time. Performance tracking and reminders work together.

Sharing tracking between members

On a shared domain, counts and yield monitoring are visible to all members: everyone can start a count and the history is built as a team, plot by plot.

Enter your actual harvests (calibration)

At the end of each campaign, enter your actual yield plot by plot. ŒnoT automatically updates the calibration of the plot (exponential smoothing), which refines the estimate for the following year.1. Open the plot sheet (web or mobile) then locate the Actual yield (after harvest) card in the right column

1. Enter the actual yield in hl/ha (e.g. 54) and the year (e.g. 2025). Previous year is pre-filled
2. Click Save. The entry appears in the history below (with a delete button if you made a mistake)
3. Calibration is applied automatically. From 2 years of history, the reliability of the estimate can change to “Calibrated” (±5%)

Target and access

The module is primarily open to independent consulting oenologists and pilot estates. To request beta access, use the “Request Beta Access” button on the section Rendement de la page d'accueil.

Methodological disclaimer: ŒnoT provides a progressive estimate of yield potential, not an exact prediction. As long as the plot has not reached the veraison stage, the confidence level is capped at “Good”. Reliability increases with field observations, counts and actual harvests entered for calibration. Any operational decision remains your responsibility.