Notes from building Herbora
Mostly about method: how we decide what counts as evidence, where the app deliberately refuses to answer, and what we get wrong. No health claims, no miracle plants.
How we grade evidence, and why four labels beat a star rating
Strong, moderate, limited, traditional use. What each one means, what it takes to move a plant between them, and why "traditional use" is a finding rather than a failure.
Read →A scan is a hypothesis, not an answer
Plant identification is probabilistic, look-alikes are common, and some of them are toxic. Why Herbora shows you a confidence score and its runners-up instead of a single confident name.
Read →When tradition and research disagree
Sometimes centuries of use meet a null trial result. Collapsing that into one verdict would be the easy design. Here is why we show you both instead.
Read →