Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

617 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

cultivOS

Precision agricultural intelligence for small and medium farms — ancestral knowledge validated with sensor data.

Tests Python FastAPI License

cultivOS turns drone imagery, soil sensors, and weather data into plain-Spanish farm advice delivered over WhatsApp. Drones capture NDVI and thermal, our intelligence layer ("Cerebro") scores field health, and farmers get a color-coded recommendation they can act on the same day — no agronomy degree required.

Not a drone company. We use third-party hardware (DJI). Our value is the intelligence layer: image processing, crop health analysis, organic treatment recommendations, and real-time field dashboards — all trained on local conditions.


Markets

Region Status Crops Farms (Y1) Primary channel
🇲🇽 Jalisco, Mexico Production Maiz, agave, berries, aguacate, cana, chile 20 WhatsApp (Spanish)
🇨🇦 Ontario, Canada Pilot Corn, soy, wheat, apple, grape, greenhouse 3 demo SMS (bilingual)

The platform is region-aware at every layer — phenology calendars, disease libraries, ancestral/traditional methods, fertilizer recommendations, and cost estimates switch based on farm location (state + country).


The Thesis

Farmers don't need to be convinced of regenerative practices philosophically — they need yield data and income math that makes it the rational choice. cultivOS makes that math visible: in WhatsApp, in plain Spanish, with color instead of numbers.

"El suelo que cuidas hoy alimenta a tus nietos mañana."


What's Built

3,800+ tests · 80+ API modules · 85+ frontend pages · 8 demo farms across 2 countries

Intelligence Layer ("Cerebro")

Module What it does
Health scoring Composite 0-100 score from NDVI + soil + thermal + trend + microbiome
NDVI processing Multispectral band math → vegetation index → zone classification
Thermal stress Water stress map + irrigation deficit detection from drone thermal imagery
Disease/pest detection NDVI anomaly patterns classified as disease vs pest vs nutrient deficiency
Crop photo analysis Upload field photos for instant visual classification
Growth stage tracking Phenology-aware development monitoring
Anomaly detection Statistical outliers flagged across time series
Data fusion Multi-sensor health synthesis with confidence scores

Recommendation Engine

Module What it does
Treatment recommendations Organic-only, region-aware, cost in MXN or CAD per hectare
Soil amendment calculator Prescribes composta, cal, azufre, etc. from soil test values
Crop rotation planner Multi-year plan based on last crop, regional seasons, soil health
Irrigation optimizer 7-day water schedule from weather + soil + thermal
Yield prediction SIAP baselines (MX) + OMAFRA baselines (ON), health-adjusted with confidence ranges
Intervention scoring Cost-benefit ranking of available treatments
Regenerative scorecard 12-point assessment of regenerative practice adoption
Carbon MRV Soil carbon measurement, reporting, verification

Knowledge Base

Module What it does
Fertilizers 13 organic methods (composta, bocashi, biochar, cover crops, wood ash, cattle manure…)
Ancestral methods 12 traditional practices — 8 Mesoamerican (milpa, chinampa, terrazas) + 4 Ontario (cover cropping, corn-soy-wheat rotation, windbreaks, companion planting)
Crop database 17 crops across Mexico + Canada with growing seasons, companions, optimal temps
Disease library 15 diseases with organic treatment protocols — regionally tagged
Regional profiles Climate zone, soil type, growing season, currency, seasonal notes per region
Seasonal calendar Phenology-based alerts (preparacion/siembra/cosecha/mantenimiento/frost_warning)

Farmer-Facing

Module What it does
Farm dashboard Spanish-first, mobile-friendly, color-coded health indicators
Field detail page Full Cerebro drill-down per field
Treatment timeline Before/after health score delta per intervention
WhatsApp/SMS alerts Dedup'd notifications with alert config per farm
Cooperative management Farmer group CRUD with aggregate dashboards
Farmer feedback Treatment trust scores from end-user ratings
Photo uploads Field photo log with ML classification

Admin & Investor

Module What it does
Intel dashboard Dark-theme admin view with cross-farm analytics
Executive dashboard Platform-wide KPIs at /ejecutivo for grant reviewers
Cerebro analytics AI decision log, accuracy tracking, activity trends
Prediction accuracy (MAPE) AI self-validation at /precision-ia — per-field and aggregate
Treatment effectiveness by crop Ranking by mean health delta
Farmer impact summary Per-farm improvement journey metrics
Portfolio report Multi-farm PDF for investors and loans
FODECIJAL report Grant-specific scientific rigor narrative
Economic impact ROI, payback, water/fertilizer savings per intervention
Regional intelligence Cross-farm aggregation by municipality and crop

Operations

Module What it does
Drone mission planning Boustrophedon waypoints from field boundary polygons
Flight log tracking Per-flight metadata, coverage, battery, outputs
Weather integration OpenWeatherMap forecasts + severe weather alerts
Soil microbiome Microbial diversity tracking with auto-classification
CSV import/export Bulk soil data import, Spanish-header farm export
PDF reports Spanish-language farm reports for FIRA loans
Role-based auth JWT with admin / researcher / farmer roles
System health Live API health dashboard for ops monitoring

Tech Stack

  • Backend: Python 3.11+ / FastAPI / SQLAlchemy / Pydantic v2
  • Frontend: Vanilla HTML/CSS/JS (no build step) + Chart.js
  • Database: SQLite (dev) / PostgreSQL (prod)
  • Processing: NumPy (NDVI band math, thermal analysis)
  • Weather: OpenWeatherMap API (global, lat/lon-based)
  • Auth: JWT with role-based access
  • Testing: pytest (3,800+ tests, region-parameterized fixtures)

Run

./run.sh
# → http://localhost:8000              Farm dashboard
# → http://localhost:8000/intel        Intelligence dashboard
# → http://localhost:8000/ejecutivo    Executive dashboard (investor KPIs)
# → http://localhost:8000/demo         Guided platform walkthrough
# → http://localhost:8000/cerebro      Cerebro AI analytics

Or manually:

PYTHONPATH="$PWD/src" uvicorn cultivos.app:create_app --factory --port 8000

Seed the demo database (5 Jalisco farms + 3 Ontario farms, 6 months of time-series data):

PYTHONPATH="$PWD/src" python3 scripts/seed_demo.py

Test

PYTHONPATH="$PWD/src" pytest tests/ -q               # Full suite (3,800+ tests, ~3 min)
PYTHONPATH="$PWD/src" pytest tests/test_ontario.py   # Ontario expansion only
PYTHONPATH="$PWD/src" pytest tests/ -k "recommend"   # Recommendation engine

Architecture

src/cultivos/
├── app.py                   # FastAPI factory (80+ routers registered)
├── config.py                # Pydantic settings from .env
├── auth.py                  # JWT + role-based access
├── middleware.py            # Error handling
├── api/                     # 80+ route modules (thin — HTTP only)
├── services/
│   ├── crop/                # ndvi, thermal, health, disease, phenology
│   ├── intelligence/        # recommendations, rotation, irrigation,
│   │                        # regions, seasonal_calendar, economics,
│   │                        # carbon, yield_model, regenerative
│   ├── drone/               # mission planning
│   ├── alerts/              # SMS/WhatsApp dispatch with dedup
│   ├── pipeline/            # CSV ingest
│   └── weather_client.py    # OpenWeatherMap wrapper
├── models/                  # 30+ Pydantic schemas
└── db/                      # SQLAlchemy ORM + session + seeds

frontend/                    # 85+ HTML pages (vanilla JS, no build)
├── index.html               # Farm dashboard
├── field.html               # Field Cerebro drill-down
├── intel.html               # Intelligence dashboard
├── ejecutivo.html           # Executive KPIs
├── cerebro-analytics.html   # AI decision log
├── precision-ia.html        # MAPE accuracy tracker
├── fotos.html               # Field photo upload
├── cooperativa.html         # Cooperative management
├── calculadora-suelo.html   # Soil amendment calculator
└── … and ~75 more

Web surfaces (three, by design)

The repo has three distinct front-end surfaces — they are not duplicates:

Path What it is Served by
frontend/ The product — the vanilla-JS dashboard app (85+ pages) FastAPI (run.sh, :8000)
root index.html + CNAME Static marketing landing for cultivosagro.com GitHub Pages
web/ Newer Next.js / TypeScript landing site (separate package.json) standalone (own deploy)

Farmer-facing pages in frontend/ use Spanish filenames (campo, cooperativa, calculadora-suelo); developer/admin pages use English (intel, field, flights). This is intentional — Spanish-first product, English-first tooling.

Region-Aware Design

The platform supports multi-country operations via a clean abstraction:

# src/cultivos/services/intelligence/regions.py
PROFILES = {
    "jalisco_mx": RegionProfile(climate="tropical_subtropical",
                                soil="andosoles volcanicos",
                                currency="MXN",
                                key_crops=["maiz", "agave", "berries"]),
    "ontario_ca": RegionProfile(climate="temperate_continental",
                                soil="glacial till",
                                currency="CAD",
                                key_crops=["corn", "soy", "wheat", "apple", "grape"]),
}

Seasonal calendars, recommendation engines, and knowledge queries all accept a region parameter. Default behavior (region="jalisco") preserves backward compatibility — adding Canada didn't break a single existing Jalisco test.


Key Metrics (targets)

  • 15–25% input cost reduction per farm (aligned with MDPI 2025 meta-analysis of 85 precision-ag studies)
  • 85 farms by Year 3, 170 by Year 5
  • 57% water wasted by irrigation inefficiency in Mexican agriculture (CONAGUA 2017–2018) — we close this gap
  • ~0% precision ag adoption on small farms in Mexico (FAO) — our TAM

Team

  • Sebastian Sanchez Garcia — CEO / Co-founder. Guadalajara-born, Toronto-based.
  • Mubeen Zulfiqar — CTO / Co-founder. MSc CS Waterloo.
  • Arshia Heravi — Software Engineer / Co-founder. Creator of AutoAgent.

Partnerships

  • ITESO (Guadalajara) — Academic validation partner. Drone lab + campus garden.

Funding

  • FODECIJAL 2026 — $2M MXN applied (Cerebro AI validation, deadline May 14, 2026)
  • Impulsora de Innovación — $6M MXN target (hardware + operations)

Roadmap

Year Stage Corridor Farms Revenue (MXN)
Y1 Launch Valles Centrales de Jalisco 20 $2.6M
Y2 Growth Deeper Jalisco penetration 45 $6.0M
Y3 Scale + Altos de Jalisco + Ontario pilot 75 $10.1M
Y4 Regional + Costa Sur + aguacate zone 110 $15.0M
Y5 Leader Full Jalisco + Ontario operational 170 $24.0M

Beyond agriculture (same drone fleet, new markets)

  • Humanitarian: water delivery to marginalized communities
  • Emergency: wildfire detection + thermal imaging
  • Environmental: water quality, reforestation monitoring
  • Commercial: solar panel and building inspection
  • Construction: progress monitoring, volumetric measurements

License

Proprietary — CultivOS Mexico S.A. de C.V. All rights reserved.

Website: cultivosagro.com Contact: partnership, investment, and pilot inquiries welcome.

About

Precision agricultural intelligence for small and medium farms — drone imagery + ancestral knowledge, validated with sensor data. WhatsApp-first, Spanish. Mexico + Canada.

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages