Verdant AI Doctor
How AI Doctor Works
AI Doctor reviews your plant history, recent logs, photos, and source-labeled sensor context — then shows its evidence, confidence, missing information, and what not to do.
Evidence first, not guesses
AI Doctor does not pretend certainty from one photo or one sensor reading. The stronger your plant memory — diary entries, photos, and source-labeled sensor context — the more useful the guidance becomes.
When context is thin, AI Doctor names what is missing instead of guessing around it. A cautious "I need more information" is a valid answer.
Sensor readings carry their source label at every step: live, manual, csv, demo, stale, or invalid. Bad or unknown telemetry is never treated as healthy.
The 12-field output contract
Every AI Doctor review returns the same structured fields, so the grower always knows what was considered and what was not.
- Summary. A short, plain-language read of what AI Doctor saw in your logs, photos, and source-labeled sensor context.
- Likely issue. The single most probable issue given the evidence. Named cautiously, not asserted as fact.
- Confidence. How strong the evidence is. Weak context produces low confidence — never a false certainty.
- Evidence. The specific logs, photos, and readings AI Doctor used, with their source labels (live, manual, csv, demo, stale, invalid).
- Missing information. What AI Doctor did not have. Named openly instead of guessed around.
- Possible causes. Alternative explanations worth ruling out before acting.
- Immediate action. A cautious next step the grower can take now. AI Doctor suggests; the grower decides.
- What not to do. Actions to avoid — especially aggressive nutrient, irrigation, or high-stress moves from weak evidence.
- 24-hour follow-up. What to check tomorrow so the grower can tell whether the plant is responding.
- 3-day recovery plan. A short, reversible plan the grower can adjust as new logs and readings come in.
- Risk level. How urgent this looks. Bad or unknown telemetry is never treated as healthy.
- Action Queue suggestion, if appropriate. When useful, AI Doctor proposes a suggestion. It reaches the approval-required Action Queue only when the grower chooses to add it, and the grower still confirms before anything is marked done.
What "missing information" means
AI Doctor surfaces the specific gaps that would make its next review more useful. Common examples:
- No recent watering log
- No pH or EC context for the most recent feed
- Stale sensor snapshot older than the freshness window
- Photo-only evidence with no environmental readings
- Unknown medium or growth stage
- Missing recent feeding information
- Unknown pot size or reservoir volume
- No stated grow target for the current stage
- No plant history to compare earlier stress or recovery
The next Quick Log or sensor snapshot the grower captures closes those gaps for the next review.
Grower-approved decisions
AI Doctor may suggest actions. The grower decides. A suggestion reaches the approval-required Action Queue only when the grower chooses to add it, and every meaningful step still waits on the grower.
Verdant does not control lights, fans, irrigation, humidifiers, or other equipment. AI Doctor cannot touch equipment. Its job is to make the next grower decision safer, not to execute it.
Plant memory. Sensor truth. Grower-approved decisions.