Corns corns.dev

corns.dev · an IoT and AI company

We use AI to automate farms.

Corns is an IoT and AI company. We put sensors and controllers on the farm, and we use AI to automate the work those readings point to: when to water, what to sow, when a tray should leave the cabinet, and when a pump should stay off. Every farm is a custom solution. The tools and technologies stay the same.

Open the live demo Design a system for your farm
  • IoT on the plot
  • AI that automates the work
  • Custom farms, same tools

The Ariana demo is one custom farm built on that platform. Same screens, sample data, nothing to install.

Dawn on a small plot: a glowing timber nursery cabinet, raised beds of lettuce and peppers, a chicken coop, and a well pump against a whitewashed wall.
Ariana demo plot 06:30 · 2.46 m in the well · lettuce ready for bed B
2.46 mWell level, a little over half of a 4.5 m column. About 2,780 liters.
23.4°CInside the cabinet, inside the tomato band of 18–26°C and 64% humidity.
31%Soil moisture on bed A. Dry enough that the morning list says water it.
6 daysLayer mash left at 43%. Sunday’s refill is already written down.
IoT on the ground.Nodes, probes, relays, a cabinet, and a camera. The farm reports itself.
AI decides the next move.The advisor ranks what to sow. Schedules shorten a run when rain is coming. Rules start the pump or open the door.
Your farm, our platform.A nursery, a well, a flock, or all of them. Custom layout. Same tools and technologies.

The company

One set of tools. A custom system for each farm.

We do not ship a different product for every customer. Corns is the platform: the devices, the safety limits, the AI, and the board. We assemble a solution for your plot from those same technologies, and we leave out the pieces you do not need.

IoT

The farm is on the network

Soil, air, well level, cabinet climate, door, mist tank, and mash level come from devices on the plot. Commands go back out to the pump, the lights, the drinkers, and the valves.

AI

Automation with a reason

AI ranks the week’s sowings, writes a pump plan from soil and rain, and checks the fence camera for a person. Each answer includes the reason. Pump safety still has to agree before a run starts.

Where the AI runs, and why
Custom

Built for this ground

Tell us the work. We fit the cabinet, the well, the beds, the flock, or a smaller mix, on the same dashboard you can open in the demo.

Where we use AI

Three jobs. Each one shows its work.

The model reads the farm you already measure: sensors, the forecast, the seed drawer, the crop, and the pump limits. It returns a recommendation with a reason. You can open that reason on the board. A pump run still has to pass the well level, the rest timer, and the daily cap, and an irrigation plan becomes a schedule when you accept it.

Corns planting advisor. This week it recommends lettuce, a half tray, with 3.4 mm of rain and two seed packets in stock. Cucumber is marked buy seeds.
Planting advisor · Ariana demo

01 · Planting

What to sow this week.

On the demo plot the answer is lettuce now, hold another tomato sowing, and skip cucumber because that packet is not in the drawer.

Open the advisor
Where
The planting screen. You ask for a cabinet tray, a field sowing, or a transplant, and you can open the score behind the ranking.
How
Every crop is scored for this month from its planting window, the forecast and the local climate, soil moisture on the node, seed you already have, free space in the cabinet, and whether the well can spare the water. A model reads that context and returns a short list: what to start, what to hold, the risk, and the next action. If the model is offline, the scored list is still the answer, with the reasons beside the score.
Why
A wall calendar does not know this week’s heat, this bed’s moisture, or which packet is actually in the drawer. The advisor is there so the sowing matches the farm in front of you, including the seed you would otherwise have to buy.
Corns AI schedule assistant. Two pump plans: an 8 minute dawn run at 78 percent confidence because soil is 31 percent, and a 4 minute afternoon pulse.
AI schedule assistant · two runs, with reasons

02 · Irrigation

How long the pump should run.

The demo week has 3.4 mm of rain on the way. That is the kind of fact that shortens a run.

Open schedules
Where
Schedules, for a chosen crop and zone. The recommendation can be saved as a real pump schedule after you accept it.
How
The board starts from the latest soil reading, the crop’s target moisture, the soil type, and the salinity of the water. It subtracts rain already in the forecast and adds the water the crop will lose to evaporation. Liters become seconds, and each run is cut so it stays inside the pump’s maximum. A model can revise those runs. Every run comes back with a start, a duration, a written reason, and a confidence.
Why
A fixed clock waters on Tuesday when the soil is already wet or a storm is a day away. The plan is there so the well is used when the crop needs the water, and left alone when it does not.
Corns camera screen. A dawn frame of the cabinet and beds, with motion score 0.4 and person detection reading not detected.
Camera · person check on the dawn frame

03 · Camera

A person at the fence, confirmed.

Dawn, morning, and dusk stills stay on the camera screen. A person is called out only after detection agrees.

Open the camera
Where
The camera screen, on the stills from the fence and on the closer frame of the beds.
How
A changed frame is the first pass. Person detection then looks at the shape. Motion in a bush is not enough, on its own, to call it a person. The screen shows detected or clear, how many, and which detector made the call.
Why
A camera that alerts on every moving leaf is a camera you stop checking. Detection is there so a person at the fence is the event that reaches you.

The same tools, on one demo farm

Ariana is one custom build. Yours can be another.

The demo plot runs four of the platform’s tools together. Each has its own water and its own clock. A Corns install can start with one of them and add the rest later, because they already share the board, the safety rules, and the AI.

Seedling trays inside a timber cabinet, with a mist nozzle and warm grow lights.
Nursery cabinet

Trays that stay in band until they are ready to leave.

A closed cabinet for the weeks between sowing and the bed. Climate, light, the door, and a mist tank of its own.

  • Tomato on the demo plot is held between 18 and 26°C, humidity 55–78%. This morning the cabinet is at 23.4°C and 64%.
  • Lights follow a clock, count dark hours, and show how much of the day’s light is already done.
  • Tray mist is a short burst from the cabinet reservoir. Field irrigation draws from its own tank.
  • You start several lines in one sowing, move a tray when it is ready, and keep the note of how the batch finished.
Open the cabinet in the demo
A concrete well head and hand pump, with a black hose running toward the beds at dawn.
Well and pump

The relay closes only when the well can spare the water.

A distance probe in the column, and a pump on a weekly clock. Safety sits in front of every start.

  • The demo well is 2.46 m down a 4.5 m column, about 2,780 liters, a little over half full.
  • Before a run: minimum level, rest since the last start, starts left today, and a maximum length.
  • You see liters used since yesterday, whether the pump is online, and the countdown while it runs.
  • When rain is on the forecast, a schedule pass can propose a shorter run than the full one.
Open the well in the demo
Raised beds of lettuce with a sensor box on a post and the nursery cabinet behind them.
Beds and the node

Soil moisture from the bed, read where the roots are.

The north node reads the soil and the air. Zones and rows say which bed is planted and which is empty.

  • This morning bed A is at 31% moisture and the air is 23.6°C. That is dry enough to water.
  • Raised-bed irrigation is its own line and its own reservoir, on a clock of its own at 18:30.
  • Crops, zones, and rows stay attached to the trays, so a batch you started inside can be found outside.
Open the sensors in the demo
Brown hens at a wooden coop and a water bucket, in warm morning light.
Chicken house

Drinkers fill at 07:00 from the flock’s own tank.

The flock has its own reservoir, its own clock, and a mash bin you can see getting low before Sunday.

  • Morning drinkers run for 12 seconds from the chicken reservoir. The well relay stays as it was.
  • Layer mash on the demo plot is at 43%, about six days left. Grit is still comfortable.
  • An empty reservoir refuses the burst. You refill from the board’s warning, with the drinkers still on their clock.
Open the chicken house in the demo

The season

From the packet in the drawer to the note you keep.

Corns treats a crop as one thread. Inventory, the cabinet, the bed, and last year’s result stay attached, so the advisor can rank a week from seed you actually have and space you actually have free.

  1. 01 · Drawer

    Know the seed.

    Packets, saved seed, and germination notes. A crop that is absent from the drawer never appears as a recommendation.

  2. 02 · Tray

    Start the batch.

    Several lines in one cabinet sowing. The tray is a production batch, with a destination already in mind.

  3. 03 · Climate

    Hold the band.

    Temperature, humidity, light, door, and mist follow the crop that is sharing the cabinet this week.

  4. 04 · Bed

    Move it out.

    The transplant queue names the row, the zone, and the day. Empty ground and full trays are visible together.

  5. 05 · Memory

    Keep the result.

    Germination, days to transplant, and strength come back the next time that seed is in season.

A person walking the path between lettuce beds toward the nursery cabinet at dawn.

A working morning

By the time you are on the path, the board has sorted the work.

The hardware reports. Safety decides what is allowed. You get the short list: water this, move that, leave the tomato tray where it is.

  1. 05:50 · the nodeSoil on bed A is 31%. The well is holding 2.46 m. Both numbers are on the overview before anyone opens a valve.
  2. 06:30 · the pumpDawn irrigation asks to run. The board checks the level, the rest timer, and how many starts are left today. Then it starts, or it tells you why it waited.
  3. 07:00 · the flockDrinkers take twelve seconds from their own tank. Mash is under half, so Sunday’s refill is already on the list.
  4. 08:10 · the cabinetClimate is inside the tomato band. Lights are counting the day. Lettuce batch 9 is flagged for bed B.
  5. 09:00 · youThe advisor says start lettuce, hold another tomato sowing, and sow basil while that seed is fresh. Cucumber is the packet still missing from the drawer.

Who it is for

Farms that want the work automated, on their own layout.

Corns is for a family plot or a market garden that wants IoT on the ground and AI choosing the next action. We design the mix around the farm you already run: a well, trays under cover, beds, a flock, or the combination that matches the work.

The technologies do not change from farm to farm. The assembly does. The board you open on corns.dev is the same one we point at your hardware.

Apply for a place See what the advisor would sow
Tomato and lettuce seedlings in trays inside the nursery cabinet.

The board

The tools a custom farm is built from.

These are working pages on the Ariana plot, and they are the same technologies we assemble for a customer. Twenty-one screens: watch the plot, grow the season with the advisor, and let AI and schedules run the water inside the safety limits.

Watch

See the plot before you walk it.

Levels, soil, weather, who is online, and the warnings that still need a person.

Overview

Well, pump, and the next window

Water left, whether the relay may close, liters used since yesterday, the cabinet at a glance, and the irrigation that is queued.

 Sensors

Soil, air, and level

Live readings and the history you would otherwise keep in a notebook. The farm and the cabinet each have their own tab. Bed A is the dry one this morning.

 Devices

Node, cabinet, camera, pump

Who is online, the last command, and the log line that explains it. The north node is the one in the beds.

 Health

The board’s own pulse

API status, database, uptime, devices seen in the last two minutes, open alerts, and how fast the board is answering.

 Weather

The week the advisor is using

Rain, heat, and the farming note for today. The demo week has 3.4 mm on the way, and one afternoon climbs above 30°C.

 Alerts

What still needs a person

Open warnings, rules you can edit, and a bell that counts what is unresolved.

 Camera

Dawn, morning, and dusk

Stills from the fence across the day, plus a closer live frame of the beds and the node.

Grow

Run the season as one pipeline.

Seed you have, trays in the cabinet, beds outside, and the result you want next year to remember.

Cabinet

Climate, light, door, mist

The tomato band, today’s light, the door, the mist tank, racks, and a way to start the next tray on more than one line.

 Seeds

What is actually in the drawer

Packets, saved seed, and germination notes, with the plant catalog to fill a name. The advisor recommends only what is on hand.

 Crops

Beds, rows, and batches

What is targeted, which row is empty, and which trays are still in the cabinet versus already outside.

 Planting

Start lettuce. Hold tomato.

The week is ranked for the cabinet, the field, or a transplant, from season, the forecast, seed on hand, and free rack space. You can open the score.

 Production

Keep the pipeline full

Sow suggestions, the transplant queue, lead time between a sowing and a harvest, and the batches that need attention.

 Learning

What last season actually did

Observed germ, days to transplant, strength, and the year-over-year note the next sowing should inherit.

 Chicken house

Water and mash for the flock

Drinkers at 07:00 from their own reservoir, and a mash bin that says when Sunday’s refill is due.

 Vegetable beds

Bed lines on their own tank

Bed drip at 18:30, from the beds’ own reservoir. A separate clock from cabinet mist and from the well pump.

Run

Water moves only when the rules say yes.

Schedules, pump safety, the well’s geometry, automation, and a pause switch that leaves the flock and the beds on their own clocks.

Schedules

Pump on a clock

Weekly irrigation, plus a pass that shortens the run when rain is coming.

 Safety

The pump waits for water

Rest between starts, a daily cap, a minimum well level, and a hard stop on how long a run may last. Automation stays inside those limits.

 Wells

A column, turned into liters

Name the well, set depth and diameter, and point the level probe at the right device. The demo column is 4.5 m deep.

 Automation

A rule, and the reason it stayed quiet

Dry soil can start the pump. A hot cabinet can open the door. You can run the set and see why nothing fired.

 Ops

Pause the plan, keep the flock watered

Stop cabinet automation, clear the pending queue, and read production attention. Chicken water and bed lines keep their clocks.

 Settings

The farm’s place, and its memory

Location for the forecast, camera display, how long readings are kept, and which cabinet this board is talking to.

A custom Corns system

We fit the platform to your ground.

Pick the jobs you want automated. We build that solution from the same IoT devices, AI, and board you see here. The pieces below are the menu, and the mix is yours.

Apply for a place

  • Nursery cabinetClimate, light, door, mist reservoir, and multi-line tray starts.
  • Well and pumpLevel probe, liters, and a relay that checks safety before it closes.
  • Bed nodeSoil and air in the beds, tied to zones, rows, and batches.
  • Bed irrigationDrip on the beds’ own tank and its own evening clock.
  • Chicken houseMorning drinkers and a mash level you can see coming due.
  • Fence cameraThe day in stills, and a closer frame when you want it.
  • The AI and the boardAdvisor, schedules, automation, and alerts, on the same login as the hardware.
  • The demo, todayOne example farm, open on corns.dev, built from these same tools.

The line

Tell us the farm. We design the system.

We are an IoT and AI company, and we are taking a short list of farms. Describe the plot and the work you want automated. We reply with a custom Corns system built from the same tools as the demo.

If you want to see the platform first, the Ariana farm is open on corns.dev. Same technologies, sample data, nothing to install.

Open the demo