Interactive Playground

Experience Agricultural Intelligence Hands-On

The FildraAI Playground is your interactive environment for exploring our agricultural intelligence ecosystem. Test image-based disease diagnosis, search our evidence-backed knowledge base, and analyze regional climate and crop data—all in one unified interface built in Africa, validated in the field, and designed to scale globally.

Explore Playground View All Products

Playground Modules

Six Modules, One Unified Experience

Each module demonstrates a core capability of the FildraAI platform. Together, they form an integrated intelligence system where vision models, knowledge retrieval, and geospatial analysis work in concert to deliver contextual, explainable agricultural guidance.

Module 01 · Map Intelligence

Live

fieldmap Map Explorer

Interactive Regional Intelligence & Climate Analytics

Explore agricultural regions with our interactive map interface. Select any country, province, or district to access climate profiles, cropping calendars, soil characteristics, and agroecological zone data. The map layer connects directly to FieldKB, surfacing location-specific knowledge so every insight is calibrated to your exact region.

Province and district-level boundary selection with real regional names
Historical climate data integration via NASA POWER (50+ years of temperature, rainfall, solar radiation)
Real-time weather forecasts via OpenWeather API (current conditions + 5-day outlook)
Agroecological zone classification with crop suitability indicators
AI-powered location insights connecting map selections to FieldKB knowledge
Multi-language support for regional content (English, Traditional Chinese, Kiswahili)

MapLibre GL NASA POWER OpenWeather AWS Bedrock

Try fieldmap Documentation

What You Can Explore

Climate Analytics

Access historical temperature, rainfall, humidity, and solar radiation data. Visualize growing degree days, drought stress indices, and disease risk periods.

Regional Profiles

View soil types, elevation, typical cropping calendars, and major pest/disease pressures for any selected region.

Map selections automatically link to FieldKB knowledge entries for that region.

Module 02 · Vision Models

Live

FieldVision AI Diagnosis

Explainable Crop Disease Detection with AI Focus Area Visualization

Upload crop photos for instant disease and pest detection powered by our DenseSwin-based computer vision models. Unlike transparent AI, FieldVision shows exactly what the model observed through AI focus area map overlays. Every diagnosis links directly to FieldKB for treatment guidance, creating a complete diagnostic-to-action workflow.

Top-3 disease predictions with confidence scores and per-class AI focus area visualizations
Crop-specific model architectures (DenseSwin for maize, ResNet for tomato, EfficientNet for cassava)
Models trained on real field images, not just controlled lab datasets
Severity estimation to help prioritize response urgency
Automatic linking to FieldKB for symptoms, treatments, and safety information
Downloadable diagnostic reports for extension officers and agronomy teams

DenseSwin Transformer AI focus areas PyTorch Field Datasets

Try FieldVision Documentation

What You'll Be Able To Do

Image Diagnosis

Upload photos of diseased leaves, stems, or fruit. Receive ranked predictions with visual explanations showing which image regions influenced the diagnosis.

Treatment Recommendations

Diagnoses automatically retrieve relevant FieldKB entries with treatment options, safety intervals, and region-specific product recommendations.

Vision results include "when not to trust AI alone" guidance for safe decision-making.

Module 03 · Voice Intelligence

Live

FieldAudio Voice Assistant

Voice-to-Voice Agricultural Advisory for Bantu Languages

Speak your farming question in Bemba, Lozi, Tumbuka, Tonga, Chichewa, Swahili, or English and receive a spoken AI agronomic answer — without typing anything. FieldAudio uses Whisper for speech recognition, Meta's NLLB-200 for Bantu language translation with a Swahili pivot strategy, Gemini for agricultural reasoning, and Meta MMS-TTS for voice synthesis in languages Google Cloud TTS doesn't support. Built for low-literacy, low-connectivity field conditions.

Whisper ASR for speech-to-text across African and international languages
NLLB-200 translation with Swahili pivot — handles Bemba, Lozi, Tumbuka, Tonga natively
Agricultural glossary preservation — crop and pest terms survive translation accurately
Meta MMS-TTS voice synthesis for languages Google TTS cannot produce
Gemini-powered agronomic reasoning with full FildraAI knowledge context
Designed for low-literacy users — no typing, no reading, just speak and listen

Whisper ASR NLLB-200 Meta MMS-TTS Gemini

Try FieldAudio

Supported Languages

Bantu Languages (NLLB + MMS)

Bemba · Lozi · Tumbuka · Tonga · Chichewa · Shona · Zulu · Xhosa · Swahili · Luganda · Kinyarwanda

International Languages

English · French · Yoruba · Hausa

Quality varies by language — your feedback shapes the next release.

Module 04 · Knowledge Engine

Live

FieldKB Knowledge Explorer

Evidence-Based Agricultural Knowledge with Clear Sources

Browse our structured knowledge base covering crops, diseases, pests, nutrient deficiencies, and management practices. Every answer traces back to cited sources—peer-reviewed research, government guidelines, and validated field data—so recommendations never arrive as a transparent AI. FieldKB powers the intelligence layer for all FildraAI products.

Tens of thousands of curated documents organized by crop, stress type, and region
Semantic search with transparent citation trails for every recommendation
Country and region-aware filtering (Zambia, Taiwan, and expanding)
Disease profiles with symptom descriptions, lifecycle information, and management options
Treatment pathways including chemical options, safety intervals (PHI/REI), and cultural practices
Severity mapping to help prioritize response actions

FAISS Vector DB Semantic Search Structured YAML RAG Pipeline

Try FieldKB Documentation

What You Can Search

Disease & Pest Profiles

Search for any crop disease or pest to access symptoms, conditions that favor spread, and recommended management actions with scientific references.

Treatment Guidance

Access chemical and cultural control options with proper safety intervals, application rates, and region-specific product availability.

All knowledge entries include source citations and licensing information.

Module 05 · Farm Memory

Beta

FieldState · Farm Memory

Your Field Today — What Needs Attention, What Has Changed

FieldState remembers every diagnosis, question, and map selection you make across FieldVision, FieldKB, and FieldMap. It synthesises that activity into two answers: what needs your attention now, and what is the current state of your field. No dashboards to interpret — FieldState interprets your farm for you.

AI-ranked attention items: urgent, watch, or stable — so you know what to act on first
Current field snapshot: crop, growth stage, location, risk level, last diagnosis — all in one view
Unified activity feed from FieldVision diagnoses, FieldKB questions, and FieldMap selections
Farm context shared across tools — set your crop and location once, every tool starts smarter
Season report with disease history, spray log, and next-season recommendations
Available to logged-in users — your field memory persists across sessions and devices

Session Memory Attention Ranking LLM Summarisation Cross-Playground Sync

What FieldState Tracks

Attention & Risk

Open spray windows, developing deficiencies, and high-confidence disease detections — ranked by urgency so you act on the right thing first.

Activity History

Every FieldVision diagnosis, FieldKB question, and FieldMap region selection logged with AI summaries — filterable by type and day.

Requires a free account. Guest users see a prompt to sign up.

Integration

How Everything Works Together

The Playground demonstrates how FildraAI's three core engines—Vision, Knowledge, and Analysis— combine to deliver contextual, explainable agricultural intelligence in a single workflow.

When you interact with the Playground, your request flows through an orchestrated pipeline. Each module adds context, validation, and explainability before results reach you. This architecture ensures that every recommendation is grounded in evidence, calibrated to your location, and transparent about its reasoning.

Step 01

Input Capture

User provides input: an image for diagnosis, a knowledge query, or a map location selection. The system identifies input type and routes to appropriate processing pipeline.

Step 02

Context Collection

System gathers contextual data: user location triggers NASA POWER climate history, OpenWeather forecast retrieval, and region-specific knowledge base routing.

Step 03

AI Processing

FieldVision runs image classification with AI focus areas. FieldKB retrieves relevant documents. ML models assess risk factors and generate predictions with confidence intervals.

Step 04

Response Assembly

Results combine into an explained response: diagnosis with visual evidence, treatment options with citations, weather-aware timing guidance, and safety recommendations.

Capabilities

What You Can Explore

The Playground supports a range of agricultural intelligence scenarios, from disease diagnosis to regional planning to treatment optimization.

Diagnosis

Identify Crop Problems

Upload an image of a diseased leaf, nutrient-stressed plant, or pest damage. Receive ranked predictions with visual explanations and direct links to treatment guidance.

Knowledge

Search Agricultural Information

Query our knowledge base for disease profiles, management practices, chemical options, and regional recommendations—all with transparent source citations.

Climate

Analyze Weather Patterns

Access historical climate data, current conditions, and forecasts for any location. Understand how weather patterns affect disease pressure and crop development.

Regional

Explore Location Intelligence

Select any region on the map to view agroecological zones, soil types, cropping calendars, and location-specific pest and disease pressures.

Planning

Evaluate Crop Suitability

Input your field location, soil characteristics, and available resources. Receive recommendations for suitable crops based on climate patterns and regional expertise.

Treatment

Get Management Guidance

From diagnosis to action: access chemical and cultural control options with proper safety intervals, application timing, and region-specific product availability.

Data Sources

Transparent Data & Licensing

Every piece of information in the Playground traces back to documented sources. We maintain clear provenance and respect licensing requirements for all data.

FieldKB Knowledge Sources

Our knowledge base is built from openly licensed and properly attributed sources, curated for accuracy and regional relevance.

CC0 and CC-BY licensed agricultural datasets
CGIAR research publications and bulletins
FAO technical documents and guidelines
Government crop calendars and extension materials
Peer-reviewed open access research papers
Regional pest and disease surveillance data

FieldVision Training Data

Our vision models are trained on properly licensed datasets and internally collected field images with clear provenance.

KaraAgro CC0 maize disease dataset (Ghana)
Harvard Dataverse African crop image collections
Internally annotated field images from Zambia operations
Partner-contributed images with data use agreements
DenseSwin architecture with AI focus areas explainability

fieldmap Data APIs

Climate, weather, and geospatial data come from authoritative public APIs and licensed mapping services.

NASA POWER API for historical agroclimate data
OpenWeather API for current conditions and forecasts
MapTiler for vector tiles and terrain visualization
Regional soil databases and elevation models
National statistical services for agricultural data

Privacy & Data Handling

We protect user data and maintain transparency about how Playground interactions are processed and stored.

Playground sessions are temporary and encrypted
Uploaded images are processed and not stored permanently
Location data is used only for contextual intelligence
No user data is shared with third parties
Explicit consent required for any persistent data collection

Important Notes

About the Playground

The Playground is designed for exploration and evaluation. Here's what you should know before diving in.

Experimental Features

These tools are in active development. We test first in African contexts, then adapt for other regions, rather than assuming a single global template works everywhere. Some features may change as we incorporate feedback.

Your Feedback Matters

We build for farmers, agronomists, and researchers. Feedback from Zambia, Taiwan, and partners worldwide guides our roadmap more than benchmarks alone. If something doesn't work as expected, we want to hear about it.

Not a Replacement for Expertise

Playground outputs support decision-making but don't replace local expertise, field observation, or professional consultation. Where AI is uncertain, we emphasize caution and recommend verification with agronomists or extension officers.

Ready to Explore More?

The Playground is just the beginning. Learn about our full product suite, discuss deployment options for your organization, or explore how FildraAI can support your farming operations, research programs, or extension work across Africa and beyond.

View All Products Contact Us