CAPABILITY/01
Scenario modeling
Test market days, target weights, and feed strategies against margin impact before committing to a plan. Side-by-side scenario comparison at the grade-mix level.
VANTIS™ Intelligence · AI-EYE™ Camera System
AI-EYE™ Camera System — Predictive Harvest Control for Processing Margin Optimization.
AI-EYE™ is the computer-vision system at the core of VANTIS™ — KYNAD's precision-poultry intelligence platform. A purpose-built camera engineered from scratch by CALYX — not an off-the-shelf camera bolted onto third-party software. Flock weighing in the house. A predictive forecast at the plant. A daily scenario for the harvest planner.
~98.4%
Real-time weight accuracy
97.5%
3-day forecast accuracy
−77%
Below-band misses · above-band −99%
Source: CALYX field data, US multi-spec integrator
Source: CALYX field data, US multi-spec integrator
10-day
Forward harvest forecast, refreshed daily
How It Works — signal in the house, decision at the gate.
AI-EYE™ is a closed loop. Cameras observe the live flock continuously. The model converts visual signal to a real-time weight curve. AI-EYE™ projects the curve forward ten days and maps it to processing-grade cutoffs. The planner sees the decision before the bird hits the line.
Accuracy Performance.
Every accuracy figure AI-EYE™ publishes is sourced from AI-EYE™ system performance — a continuously-running validation suite that compares model output against scale-recorded ground truth, validated with major U.S. poultry integrators on a platform with information security controls aligned with ISO/IEC 27001.
The validation suite
-
Real-time weight · ~98.4% accuracy (1.6% MAPE) Validated against plant weights on more than 1.4M birds, across multiple breeds and more than 7 countries. Range bracketed by complex and genetics; figure shown is the steady-state mean.
-
10-day supply forecast Forward weight prediction runs to F+7 — 97.5% at 3 days, 96.7% at 7 days. The 10-day view is a supply and harvest-planning horizon built from those predictions plus known lead times, not a weight-accuracy figure.
-
In-spec supply · below-band misses −77% A US multi-spec integrator tightened below-band misses 77% and above-band misses 99%, with the weekly forecast matching plant actuals within 1%. A US single-spec complex predicts the heavy tail before harvest, keeping oversized product under the divert limit. Source: CALYX 2026 Brochure, field results.
Why accuracy ranges, not point claims. Performance varies by complex, genetics, and house density. We publish steady-state means with the range visible — never a best-house figure dressed up as a guarantee.
AI-EYE™ · Complex 04/House 12
Current weight
2.18kg
↑ on curve · +1.2%
Forecast · D+10
2.58kg
Premium band
Margin impact
+$0.14/bird
vs. plan baseline
10-day forecast · uniformity-weighted
Premium band 2.45–2.70 kg
illustrative house view
Below-band misses −77% · above-band −99% (CALYX field data)
Sources: CALYX 2026 Brochure (FINAL v3) — real-time weight accuracy and field results. CALYX AI Camera Progress, Global Performance Summary — 3-day and 7-day forward prediction. Real-time weight accuracy is ~98.4% (1.6% MAPE), validated vs plant weights on more than 1.4M birds. Forward weight prediction is 97.5% at 3 days and 96.7% at 7 days; the 10-day view is a supply and harvest-planning horizon, not an accuracy figure. CALYX monitors ~84M birds annually with 800+ cameras (~14M birds per cycle).
Harvest Intelligence Software — data, modeling, and the planner's interface.
The platform integrates the modeling and reporting layer with the AI-EYE™ vision system. It ingests house-level weight signal, runs the forecast model, and presents the planner with scenario impact at the grade-mix level. Built for integrator-grade rollouts.
CAPABILITY/02
Multi-complex dashboard
Roll up from individual house to complex to integrator level. Deviation alerts surface when forecast drifts outside the planner's stated band. Per-grade contribution to margin visible at every level.
CAPABILITY/03
Daily-refresh forecast
The 10-day forecast refreshes daily as new weight signal arrives. Late-cycle confidence band tightens; early deviations are flagged when they would compound through to harvest.
Deployment & Integration.
AI-EYE™ deploys at the complex level — cameras in-house, model in the cloud, data integration into the planner's existing tools. KYNAD services handle the install and on-board the live-production and planning teams.
| Step | What happens | Owner | Typical timing |
|---|---|---|---|
| 01 · Site survey | House-by-house assessment of camera placement, network coverage, and existing weighing infrastructure. | KYNAD field engineering | Week 1–2 |
| 02 · Install | Camera mount, power, and network drop in each participating house. Calibration against scale-recorded ground truth. | KYNAD + integrator ops | Week 3–4 |
| 03 · Data integration | API or file-drop integration with the integrator's planning system, ERP, and grade-reporting infrastructure. | KYNAD platform + integrator IT | Week 4–6 |
| 04 · Onboarding | Planner, live-production, and complex-manager training. Scenario-model walkthroughs against historical data. | KYNAD customer success | Week 5–7 |
| 05 · Live operations | Continuous monitoring, model drift alerts, monthly performance reviews against scale ground truth. | KYNAD platform + integrator | Week 8+ |
Final deployment timing depends on house count, network readiness, and existing data infrastructure. Share your operation profile and we'll return a deployment plan with specific milestones.
The margin mechanism — small shifts, large outcomes.
AI-EYE™ is a processing-margin control system: tightening flock uniformity and concentrating birds in the target weight band converts biological variability into measurable yield and pricing gains. The distribution below is the documented before/after from the harvest-intelligence dataset.
Before · high variability/low uniformity
~45–50%
of flock in the target weight band
Underweight−8% of flock
Overweight+7% of flock
After · harvest selection optimized on AI-EYE™ forecasts
~60–65%
of flock in the target weight band
Underweight−3% of flock
Shift into target band+16%
+$0.30/kg
Margin uplift from premium weight capture
+10%
Processing yield improvement
Weight-distribution and margin figures reflect documented harvest-intelligence outcomes; actual results vary by operation type, genetics, market, and feed program.
Illustrative ROI — per house.
A generic AI-EYE™ + Harvest Intelligence model for a representative large-scale operation, at KYNAD list pricing. All figures illustrative — actual outcomes vary by operation, market conditions, and grade premiums. KYNAD does not guarantee uplift or returns.
Illustrative ROI Model — VANTIS™. (45,000 birds/house · 5.5 cycles/yr · 2.5 kg avg · US$1.44/kg live · 2 AI-EYE™ cameras + AI Harvest Pro per house.)
~$32K
Value per house per year
~5.3×
Illustrative ROI
~2.3 mo
Payback
Value drivers: harvest-weight accuracy (~40 g/bird), uniformity (~1.5%) and processing optimization (~0.5%). See the full model on Harvest Intelligence Software.
Outcome
See AI-EYE™ against your operation.
30-minute walkthrough of the platform against your processing volume, grade mix, and weekly economics. We'll bring the scenario model already populated.