FildraAI began from a personal field problem, but it is being built by a growing team across research, design, markets, and field validation. The work is simple to state and hard to earn: create agricultural intelligence farmers can understand, question, and trust.
What we're fighting
Forgotten seasons. Unrecorded knowledge. Advice without reasons. Decisions made with no memory of what worked before. That is the gap FildraAI exists to close.
Our mission
Every season should teach the next.
The Problem
A single disease outbreak can threaten a family's entire livelihood. An incorrect treatment recommendation can damage crops, soil, and ecosystems for seasons to come.
Yet many digital agricultural tools are built far from the realities of the fields they aim to serve. They are developed without local context, tested without local conditions, and deployed without regard for the complexity of local agricultural systems.
In agriculture, these gaps are not just technical problems, they are matters of trust. And trust, once broken in a farmer's field, is very hard to rebuild.
Generic AI systems often provide advice that ignores local regulations, regional product availability, environmental context, and offers no explanation for how a decision was reached.
A Journey Across Two Agricultural Worlds
FildraAI's founding story spans two very different agricultural environments, each contributing something essential to what the platform became.
Zambia · Southern Province
Founder Chiuzu Chilumbu was raised in Zambia, where maize is not just a crop, it is a cornerstone of food security and daily life. Farmers across many regions still rely on manual inspection to detect crop diseases. While this knowledge is invaluable, diagnosing disease across large fields or at early growth stages is an extraordinary challenge that AI can meaningfully support.
Taiwan · Agricultural Region
During his undergraduate years at Tunghai University, Chiuzu worked on the university farm and visited Yunlin, one of Taiwan's most productive agricultural regions, gaining firsthand exposure to precision farming. He then pursued a master's degree in AI and computer vision at National Tsing Hua University, where deep learning research gave shape to the vision behind FildraAI.
The Origin
Chiuzu founded FildraAI after years of moving between farming environments in Zambia and AI research in Taiwan. That origin still matters, but the company is no longer framed as one person's story. It is a shared effort to turn field knowledge, design, engineering, and market understanding into tools farmers can actually use.
On the team, Chiuzu's clearest role is research leadership: keeping the technical work grounded in evidence, crop context, and explainability. The wider FildraAI team carries the rest of the mission: product clarity, regional insight, field validation, and the patience needed to build trust.
We keep the founder note intentionally modest because the product has to stand on something stronger than biography: useful work, careful validation, and respect for the people using it.
These are where we have ground truth today. Across Africa and Asia, farmers face the same gap, important decisions made without memory or explanation. That is where the work is going next. We will name new countries only when we have ground truth there, not before.
Our Team
FildraAI brings together research, design, regional analysis, and field operations. The point is not to center one founder. The point is to build a practical agricultural intelligence company with enough perspectives to serve farmers responsibly.
Founder & Research Lead
Leads the AI research direction, model validation, and the field-first product thesis behind FildraAI.
Education
M.S. Computer Science & AI National Tsing Hua University, Taiwan
Experience
ML/AI Research & Development Engineer Author, DenseSwin (IEEE 2024) · Field validation across Zambia & Taiwan
Graphics & Design
Shapes the visual language, product presentation, and communication assets that make the platform easier to understand.
Education
M.A. Visual Communication Design National Taiwan University of Education
Experience
3+ years in brand & product design
Asia Market Analyst
Tracks market needs, partner opportunities, and regional signals across Asia to keep product strategy grounded.
Education
M.S. International Business and Political Science National Chengchi University
Experience
5+ Asia-Pacific market research & partnerships Firm grasp of regional dynamics
Africa Market Analyst
Supports African market research, farmer context, and validation priorities for practical deployment.
Education
B.Sc. business and Finance University of Lusaka
Experience
Smallholder programs & field research, Zambia Extensive experience in agricultural development
Field Validation Work
"Utilizing a DenseSwin Transformer Model for the Classification of Maize Plant Pathology in Early and Late Growth Stages: A Case Study of Its Utilization Among Zambian Farmers"
IEEE PublicationA hybrid model combining densely-connected convolutional blocks with a shifted-window transformer attention mechanism, designed to detect subtle disease patterns at both early and later infection stages.
The Research
Chiuzu's master's research explored a direct question: could deep learning meaningfully improve how Zambian farmers detect crop disease in maize, the country's most critical staple crop?
The research introduced DenseSwin, a novel architecture that fused the spatial detail-capture of dense convolutional networks with the long-range pattern recognition of transformer models. The result was a system capable of identifying disease symptoms at both early and late growth stages, an especially critical distinction for farmers who need to act before visible damage spreads.
Through rigorous experimentation, the model achieved 97.18% accuracy in maize disease classification, results that demonstrated AI could be a genuinely useful diagnostic partner for farmers in real field conditions.
But the research also revealed something equally important, a truth that would define the entire direction of FildraAI.
Beyond Accuracy
Agricultural decisions do not happen in isolation. A farmer's field is shaped by an intricate web of regional, regulatory, and cultural factors that generic AI systems routinely ignore.
Many existing digital tools provide recommendations without verifying whether suggested treatments are available locally, legally permitted, or safe under local agricultural regulations. They don't account for Pre-Harvest Intervals or Re-Entry Intervals. They have no concept of district-level climate variation.
And most critically: they operate as transparent AI, offering predictions with no explanation of how the decision was reached.
In agriculture, where livelihoods and ecosystems are at stake, this lack of transparency can undermine trust. A farmer who cannot understand why an AI recommended a particular treatment cannot make an informed decision, and may rightfully choose to ignore the advice entirely.
What Generic AI Misses
What We Built
FildraAI was founded to move beyond isolated machine learning models toward a broader vision: agricultural intelligence that is accountable, explainable, and grounded in local context. Rather than replacing agronomists or extension services, our goal is to augment human expertise and support better-informed decisions in the field.
Crop disease detection with AI focus area explanations, the AI shows exactly what it sees.
Structured, country-specific agricultural knowledge with citations to peer-reviewed sources.
Built-in PHI/REI compliance calculations, regulatory checks, and local product verification.
District and province-level localization for every recommendation, because geography matters.
English, Chinese, Kiswahili, and French, built so language is never a barrier to safe decisions.
Every AI output is transparent, not just "what" but "why," so farmers can make truly informed choices.
A Cross-Regional Approach
Today, FildraAI operates across two complementary environments. This dual perspective ensures that our systems are not built in isolation, but developed with continuous feedback from the communities they aim to support.
Taiwan · Research & Development
Our R&D work is rooted in Taiwan's strong technological ecosystem, combining academic research foundations with access to cutting-edge machine learning infrastructure, enabling rapid experimentation and model development.
Zambia · Field Validation Operations
Our field validation work focuses on agricultural regions in Africa, ensuring our systems are not just theoretically sound, but practically trusted. Real farmers. Real fields. Real feedback that shapes every iteration of the platform.
FildraAI is entering its Field Validation Phase, continuously refining our systems through real-world use and collaboration with farmers, researchers, and agricultural institutions. We believe trust in agricultural AI is not declared. It is earned.