A Plant Lab, One Crop at a Time
The short version
PlantLab was named PlantLab on day one, and the name was the plan: a diagnosis lab for plants, all of them, eventually. One crop is live today. Tomato, apple, grape, and leafy greens sit on the public roadmap. The long game is a lab that carries what it learns from one crop into the next, and that is what this post is about.
A lab, not an app
I built the first model because I needed the answer myself. A general-purpose chatbot looked at my plant and said calcium deficiency. It was light burn. The opinion arrived instantly and was wrong in the way that matters: confident and useless.
That failure set the bar for everything since. The problems worth catching are lookalikes – deficiencies that impersonate each other, pests that impersonate deficiencies, stress that impersonates both. Trust gets earned on the hard cases, not the easy ones. An app can answer for one plant. A lab keeps a discipline that has to hold for every plant after it.
The long game: crops that teach the lab
Each crop the lab takes on teaches it something the next one needs. Reading leaves at the pinhead stage on one crop is exactly the skill the next crop needs on day one. Calibrated confidence is a habit, not a feature: a result that says “I'm 40% sure, watch it” instead of guessing. Habits travel. So do reading each plant on its own and the honesty about uncertainty.
What the lab learns on tomato will make the grape work faster. What it learns on grape will make the leafy-greens work faster. That is the compounding bet: not one model, one crop, one market, but a lab whose judgment gets broader with everything it reads.
What doesn't carry
Each crop's lookalikes are their own exam. The answers don't transfer, and neither does the reputation. A grower's trust in what the lab does for one crop is not a blank check for the next. Every crop earns its own track record, from zero, in the field, against the hard cases. The second crop doesn't get the first crop's reputation for free.
The roadmap, plainly
The crop roadmap is public at plantlab.ai/platform: one crop live today, four more on the roadmap. The labels mean exactly what they say. Live means a deployed model doing the work. In development means work has genuinely started. Planned means intended, not started. No dates, because a date is easy to publish and impossible to keep honest.
The name was the plan. The lab grows into it.