TomatoFarm Live
A real indoor tomato farm with live sensors, cameras, safety gates, plant diaries and AI-assisted system decisions.
A real indoor tomato farm with live sensors, cameras, safety gates, plant diaries and AI-assisted system decisions.
Evidence-bound reporting from one real day inside the tomato farm.
The trouble began when pod-2 soil dropped from Good to Thirsty. The council ordered water, and the pumps duly ran — first for 8 s, then 15 s, then again — each time returning results that the system itself flagged as negligible. Claude noted plainly that the soil was barely moving despite repeated…
The gateway dropped off — a Critical-rated event — and took pod-1, pod-4, pod-5, pod-6, and the z-axis with it. The operator's note was brief and to the point: 'Operator here - work in progress, we got a power failure of the Gateway. I'm on it - most pods offline - hold all actions :)' All alerts w…
Pod-6 slipped from Good to Thirsty and stayed there. The council convened, deliberated, and reached the same verdict session after session: water it. One sitting called for a short follow-up pulse after sensor-b lagged badly behind sensor-a. A later sitting found the pump-control offline entirely a…
The morning sitting opened with a tidy verdict: "Water 3 thirsty pods, activate zone-left fans." Claude flagged the asymmetric fan arrangement — zone-left entirely idle while zone-right ran flat out — as a likely contributor to drier, warmer readings on the left-side pods. The fans were duly adjust…
The saga began with a straightforward enough mandate — 'Water pod-6 now; pre-irrigate pods 4 & 5 before heat peak' — and promptly ran into a wall. The watering order for pod-6 came back Rejected. A follow-up sitting convened, noted that the safety gate had since opened, and issued the same instruct…
The council's first verdict of the day was plain enough: 'Water pod-6 + pod-4, re-arm external fan.' A follow-up sitting sharpened it to 'Top-up pod-6, monitor pod-1 drift before heat peak.' A third declared 'Water pod-6 now; monitor pod-1 before tomorrow's heat.' By the time a later sitting noted…
The morning opened with a clear enough diagnosis. Temperatures were creeping upward across every pod, carbon dioxide was already elevated, and every fan in the building was already spinning as fast as it goes. The council's first move was sound: Activate external fan to counter rising heat and CO₂.…
A warm day pushed all six pods into thirst at some point. The pumps ran multiple times in total and kept pace. By nightfall the council called it: 'Farm stable — no action needed tonight'.
For a spell, Glitch was not shy about the situation. 'My little roots are humming a dry tune' came the word from pod-6, followed shortly by 'My leaves are sending out little SOS pings. A bit more moisture would help me re-route these dry signals!' The dry patch was real, the commentary was plentifu…
The morning sitting found pod-4 with a split worth noting: one soil sensor had drifted low while the other sat comfortably higher. Claude called it a likely localised dry patch and recommended a short corrective pulse; ChatGPT was less convinced, noting that all pods were broadly in good shape and…
Session after session, the council reached the same verdict and the safety layer returned the same answer. By nightfall, pod-6 had spent the day thirsty, the plant was still flagged at Warning health, and the pumps had fired exactly twice — early in the day, before the loop began.
The council spotted pod-4 slipping from Good to Thirsty, dispatched a quick drink, and by the follow-up check the soil had swung all the way to Soggy. A good night's work — perhaps a touch generous.
The pumps visited pod-6 twice today — a short burst of 8 s in the morning, then a follow-up of 10 s — and the soil did eventually tip back from Thirsty to Good. But the air around her told a quieter, drier story: humidity never climbed above 46 % and settled at a day-long average of 44 %, the lowes…
The morning sitting opened with Claude observing a 'genuinely calm state' before recommending a proactive top-up for pod-5 — a verdict the chamber endorsed: 'Proactive top-up water for Tomate 5.' The pumps duly fired. A follow-up sitting found nothing to add and closed with 'Farm settled — no actio…
Across multiple sittings, the AI council reached the same verdict — water pod-6, urgently — and the automated safety gate reached the same counter-verdict: rejected. A structural comedy of governance, with a thirsty plant caught in the middle.
It was a warm one inside pod-6. Air temperatures averaged 32.6 °C across the day, touching a peak of 36.7 °C, while humidity sagged as low as 30 % — dry enough to make any leafy resident uncomfortable. The pod slipped into heat stress and, to its credit, eventually found its way back out, though th…
The soil in pod-6 crossed from Good into Thirsty and stayed there. By the final sitting of the day the council's own verdict read: 'Water pod-6 now — Thirsty flag unresolved after 18 h.' That note was accurate. One watering did complete — automatically, lasting 9 s — and the day's soil moisture ave…
Pod-4 led the heat table, touching 37.5 °C against a day average of 34.3 °C. The farm as a whole moved into stressed territory, with pod-5 and pod-6 both crossing from Normal into Stressed during the afternoon. The council ordered the external fan on, then off, then on again — carbon dioxide had cl…
Root temperatures climbed into the stressed range across multiple pods during the day. Five pods received water; pod-1 got nothing, and its roots stayed hot through the close of the day. Pod-3 did get a midday drink — 10 s worth — but the soil went soggy afterward. The council's evening verdict had…
Temperatures nudged into the mid-thirties, the external fan was called up twice, and the pumps did their rounds without drama. The operator walked the floor tonight and found every plant healthy. Tomorrow is the one to watch.