Not Just Answers. More Reliable Decisions.
Gardener Help is a project dedicated to more reliable understanding of plant problems and more informed decisions about plant care.
Behind the project is the idea that the right answer doesn't necessarily start with recognizing the visible symptom. It starts with understanding the problem, evaluating the available evidence, and considering the possible explanations.
Yellowing leaves, slowed growth, or spots can have different causes. Similar symptoms don't necessarily mean the same origin. That's why a hasty conclusion can lead to an inappropriate intervention, unnecessary costs, or additional damage.
Our goal is to help people move from the visible symptom to a better-reasoned decision.
Who Is Behind Gardener Help?
Gardener Help was created by Mladen Dichev — an entrepreneur and the project's developer, who works on applying artificial intelligence to practical problems in agriculture.
For more than two years now, his work has focused on building and refining a structured approach to analyzing plant problems.
This includes developing rules for evaluating symptoms, comparing competing hypotheses, determining the evidence needed, and recognizing situations where the available information doesn't allow for a reliable conclusion.
The project has grown through systematic work on the diagnostic logic, case review, and deliberate refinement of the rules used to analyze different situations.
This isn't simply using an off-the-shelf AI tool. A substantial part of the work is building the methodology that guides the analysis and determines how the system should treat evidence, uncertainty, and the risk of error.
The Methodology Behind the AI
In Gardener Help, artificial intelligence is a tool for applying a structured methodology. It is not the methodology itself.
The framework is built and refined by defining rules, criteria, and analysis sequences. The model uses these instructions to examine specific cases, evaluate possible explanations, and formulate answers.
Improving the system involves reviewing results and making deliberate changes to the logic whenever weaknesses or opportunities for more reliable analysis are identified.
This approach matters because simply using artificial intelligence doesn't guarantee a correct result. Quality depends on how the analysis is organized, on the available data, and on the system's ability to account for its own limitations.
That's why we focus on five principles:
- Evidence Before Conclusions
We distinguish observed signs from their interpretation and from hypotheses about the possible cause. - Comparing Alternatives
We consider the plausible explanations, instead of automatically accepting the first one that seems to fit. - Finding the Limiting Factor
The visible symptom doesn't always show the process that is actually limiting the plant's development. That's why the analysis must also account for other possible factors. - Acknowledging Uncertainty
When the evidence isn't sufficient, the system must reflect that limitation, instead of creating an impression of certainty. - Risk-Aware Decisions
The goal isn't simply to suggest an action, but to assess how well-supported it is by the available information and what risks it may carry.
From Gardener Help to Veltera Nova
Gardener Help is the practical domain where we develop and apply this approach.
Agriculture is a particularly fitting environment for working on decision reliability. Biological systems are variable, causes often interact, feedback can be delayed, and decisions sometimes have to be made with incomplete information.
These challenges raise a broader question: how do you build systems that don't just offer answers, but also assess the grounds for them, recognize competing explanations, and account for the limits of their own interpretation?
That's the direction of Veltera Nova — the broader initiative behind developing approaches to reliable decision-making and AI-based reasoning systems.
Agriculture is our first practical domain of application. Our broader goal is to develop and validate methods that can support more reliable decisions in other complex environments as well.
What We Want to Achieve
We want Gardener Help to help people better understand the problems in growing plants, weigh the possible causes, and choose more informed actions.
We don't claim that every situation can be resolved from a photo alone, or that an AI system can replace all the necessary measurements and professional assessment.
Instead, we work toward an approach that accounts for the real complexity of plant problems and puts the quality of the decision ahead of the certainty of the answer.
To us, a good system isn't one that always looks certain. It's one that has clear grounds for its conclusions and recognizes when the evidence isn't enough.
Connect With Us
Follow Gardener Help on Facebook: Gardener Help — Помощ за вашата градина
Connect with Mladen Dichev on LinkedIn: linkedin.com/in/mladen-dichev-75aab4402