Your timer knows what day it is. It has no idea that four inches down — where the roots are — your soil sits at 18% and falling. One stake reads that depth and turns it into a sentence: water tonight, 32 minutes.
$4.99 / device / monthCancel anytime Wi-Fi, MQTT or LoRaWANNo hub required
Reference ET₀ computed from your latitude and the day of year, then a grass crop coefficient. Named so an agronomist can check our work.
Wi-Fi over signed HTTPS, MQTT to a broker you control, or LoRaWAN through The Things Network. Chosen per device, changeable later.
Nothing else to buy, nothing else to power, nothing to wire into your sprinkler valves. The stake talks to the internet on its own.
CSV per sensor in display or raw units, up to two years and 400,000 rows, plus a full account export in JSON. Your data leaves whenever you want it to.
$4.99 per device per month from the billing portal — no phone call, no retention script. Deactivate one sensor and the quantity simply drops by one.
The problem
A watering schedule is a guess you make in April and stop questioning in July. It runs on Monday because it is Monday. Here is one ordinary week, drawn against what the soil was actually doing underneath it.
Scroll the chart sideways
The arithmetic
A worked example, not a measured result. Your heads, pressure and run time will differ — the point is that the multiplier is a schedule you set once.
What 240 gallons looks like
240GAL
Roughly three full bathtubs at about 80 gallons each, poured onto grass that was already wet — in a single week, from a single zone.
The part nobody measures
A controller reports a completed cycle whether the water reached the roots, ran down the driveway, or came out of a head the mower knocked sideways in June.
The failure mode of open-loop irrigation
0.62 in fell on Monday afternoon — eight hours after the sprinklers finished, two days before they were due to run again. The controller never found out.
The top inch dries in hours; the root zone holds water for days. Judging by the surface gets you both mistakes — watering soil that is already wet, and skipping soil that is not.
On a slope, on compacted soil, or through hydrophobic thatch, water sheets off toward the sidewalk. Our infiltration tool measures the moisture jump after a cycle, so you find out.
It has been watering the driveway since June. The timer reports success every single time. The soil reports 11%.
Fig. 01 — One week on a fixed Mon/Wed/Fri schedule, 500 sq ft spray zone. The same controller over-waters on Wednesday and Friday, then leaves the root zone below the refill point for a full day once the heat arrives. A schedule manages to be wrong in both directions in the same week.
How it works
No hub to plug in, no valve to rewire, no controller to replace. The stake measures; you decide what to do with the answer.
01 — about two minutes
Pick the spot that dries first — the sunny corner, the strip by the driveway — and push until the collar meets the thatch. No trenching, no wiring, no digging up the yard. Keep the sensing section in the root zone rather than in the top inch of thatch, which dries far faster than the roots do.
02 — every 5–15 minutes
Pick a radio at setup and change your mind later. Wi-Fi posts HMAC-signed JSON to
/api/ingest. MQTT publishes to a topic your device alone can write. LoRaWAN goes
through a Things Network gateway that may be a mile away.
Soil moisture, soil temperature, air temperature, humidity, light, rainfall, battery and signal — eight channels, every uplink.
03 — when it matters, not daily
The advisor takes today's ET₀, your soil's actual depletion and the rain your own gauge recorded, then gives you a start time and a run time. Follow it or ignore it — the reasoning is always on screen, so you can argue with it.
It is deliberately conservative. If the soil is already in the band, or rain has covered the demand, it tells you not to water.
Fig. 02 — Install, uplink, decide. The only thing that changes about your irrigation is when you turn it on.
The same week, with a sensor
Same weather, same lawn, same sprinklers. This is the root zone at four inches for seven days, with every decision written onto the moment it was made — including the two we got to before you did.
Scroll the chart sideways
Fig. 03 — Seven days of measured root-zone moisture plus a 26-hour forecast tail. Two cycles, not three; the one that ran, ran longer and went deeper. The reasoning behind each annotation is on screen in the app, next to the number that produced it.
The hardware
A sealed weather head above the turf line, two capacitive tines below it, and a radio that talks straight to the internet. The parts of this drawing that are still being finalised say so.
Battery life, ingress rating, range and dimensions are not confirmed on the shipping unit yet, so they are not printed here. We would rather show you an empty field than a number we would have to walk back.
Fig. 04 — Cutaway, not to scale. Depths on the ruler are the ones the software cares about; the tine length itself is still to be confirmed.
The dashboard
A health score that names its own limiting factor, the week's curve, and an advisor that shows the arithmetic. This is the layout, with illustrative data from the week above.
Lawn health score
Limiting factor:
root-zone moisture
Right now
%VWC at 4 in · inches applied · Mon 22 – Sun 28 Jun
Field capacity 28% · refill point 20% · permanent wilting 12% for this soil texture. The advisor aims to keep you inside the band.
What the advisor did this week
Smart watering advisor
Tonight, 9:20 pm
32 min · 0.62 in
Root zone 17.4%, 2.6 points under refill.
ET₀ 0.29 in today (Hargreaves).
Rain probability, next 12h: 5%.
A late start beats evaporation loss.
Dry-down forecast
1.4 days
from here to the stress line at 14% if you skip tonight — Tue ≈ 6 am.
This week
3 cycles skipped
360 gal not applied. One deeper 31-minute soak added Thursday night. Net −174 gal.
Anomalies
None open
Watching for leaks, dropouts, frozen probes and dying batteries.
Fig. 05 — The device page, with illustrative data. Every number on it is derived from the eight channels the stake reports; nothing is estimated from a regional weather feed.
The tools
Every one of these is derived from the same eight channels — nothing here is a separate sensor or a separate purchase. What follows is the input list for each, so you can decide for yourself whether the output means anything.
A single 0–100 number, and — more usefully — the name of the one input dragging it down.
Readsroot-zone moisture band · moisture stability · heat and cold stress · light · data freshnessGives you78 / 100 · limiting factor: root-zone moisture
Turns today’s evapotranspiration and your soil’s actual depletion into a start time and a run time.
ReadsFAO-56 Hargreaves ET₀ from your latitude and day of year · grass crop coefficient · your own rain gauge, last 24h · current depletionGives you“Tonight 9:20 pm · 32 min · 0.62 in”
Projects your moisture curve forward and says when it crosses the refill point — so you can water before it, not after.
Readsthe last several days of your own measured dry-down rate · forecast temperature rangeGives you1.4 days to refill · stress line Tue ≈ 6 am
Moisture climbing when no rain fell and no cycle ran is a stuck valve or a burst line. It also catches frozen probes, dropouts and physically implausible readings.
Readssoil-moisture slope · rainfall summed across the whole window · logged irrigation · reading cadence · battery voltageGives youopen anomalies, with the evidence that raised them
How much of what you applied actually reached the root zone, measured from the moisture jump after a cycle.
Readsmoisture before and after each cycle · applied depth · time to peakGives youapplied vs infiltrated, per cycle
Gallons and dollars against the fixed schedule you used to run — arithmetic from your own numbers, not a promise.
Readscycles skipped · cycles added · your zone’s applied depth · the rate on your own billGives yougallons and dollars, week by week
Growth is driven by soil temperature and water, so the calendar is the wrong tool for all three.
Readssoil temperature at root depth · accumulated growing degree days · moisture bandGives you“Soil 74°F — good window to seed”
Zone against zone, or this July against last July. Averaging two zones hides the problem you are trying to find.
Readsany two or more sensors, or one sensor across two periodsGives youaligned curves and the gap between them
Fig. 06 — The eight tools and their inputs. All eight are included at both price tiers; there is no plan where the useful ones are held back.
Connectivity
This is the part most soil sensors get wrong: they make you buy their hub, put it in range of their radio, and keep it powered forever. We never require one. If the lawn is in Wi-Fi range, use Wi-Fi. If it is a paddock, use LoRaWAN.
| Wi-Fi (HTTPS) | MQTT | LoRaWAN | |
|---|---|---|---|
| Pick it when | The lawn is inside your house Wi-Fi. Simplest possible setup, nothing extra to run. | You already run a broker or a home-automation stack and want the sensor on the same fabric. | A paddock, an allotment, a far corner, a rental. Range instead of bandwidth. |
| What it needs | Router coverage at the spot the stake will live. Check with a phone before committing. | A broker you control. We issue per-device credentials locked to that device's own topic. | A Things Network gateway in range — often one that is already there and not yours. |
| How it authenticates | HMAC signature plus a timestamp. A reading cannot be forged, or replayed after five minutes. | Per-device broker credentials. One device cannot read or write another's topic. | Network-session keys, plus our payload formatter on the Things Network side. |
| What it costs you | Nothing beyond the device. | Nothing beyond the device, if the broker is already running. | Nothing beyond the device, if a public gateway covers you. |
| Can you change later | Yes — the choice is per device, and a mixed fleet on one account is fine. | ||
Scroll the table sideways
Readings buffered on the device are accepted when it reconnects, and a late reading will not rewind your last-seen time. If nothing arrives for three hours we raise an offline alert and email you — and the Lawn Health Score down-weights stale data, so a dead sensor never quietly reports a healthy lawn.
Fig. 07 — All three routes end at the same validated, de-duplicated ingest pipeline. Nothing in this diagram is a product we sell you twice.
Do the arithmetic yourself
This is arithmetic on numbers you supply, not a prediction and not a promise. Every assumption it makes is printed underneath, where you can disagree with it.
— gallons a season, not applied
— at the rate you entered
A 26-week irrigation season. Distribution uniformity of 75%, meaning a quarter of what leaves the heads never usefully reaches the root zone. 0.623 gallons per square foot per inch of depth — that one is geometry, not an assumption. Saving is the difference between what your schedule applies and what the turf uses; where your schedule already applies less than the turf uses, the answer is zero, not a negative. Rainfall is excluded entirely, which makes this a conservative figure rather than a flattering one.
Fig. 08 — A model, not a measurement. Once a stake is in the ground the same figure is computed from your actual applied depth and your actual dry-down, and then it is a measurement.
Pricing
There is no plan where the agronomy is held back. The hardware is a separate one-time purchase; the subscription is what keeps your device streaming and your history alive.
Monthly
$4.99 / device / month
Cancel any month from the billing portal. Deactivate one sensor and the quantity drops by one on the next invoice.
Yearly
$49 / device / year
Works out at about two months free. Same features, same tools, same export.
Straight answers
Nothing here is marketing copy dressed up as a question. Where the honest answer is “no”, it says no.
One per area that is watered differently or behaves differently. A single front lawn on one sprinkler zone is one sensor. A shady strip that stays damp for days wants its own, because averaging the two hides the problem you are trying to find.
No, and that is deliberate. We tell you when to water and for how long; you or your existing controller do the watering. That keeps it compatible with any irrigation setup, including a hose and a mechanical timer, instead of only working with one brand of smart valve.
A soil sensor that can reach the internet by HTTPS over Wi-Fi, by MQTT, or by LoRaWAN through The Things Network. There is no proprietary hub to buy and nothing to wire into your sprinkler system. Claiming a device hands you the ingest endpoint, the request-signing snippet, an MQTT example and the full Things Network payload formatter.
Readings buffered on the device are accepted when it reconnects, and a late reading will not rewind your last-seen time. If nothing arrives for three hours we raise an offline alert and email you. The Lawn Health Score down-weights stale data, so a dead sensor never quietly reports a healthy lawn.
Yes — CSV per sensor, in your display units or the raw canonical ones, up to two years and four hundred thousand rows per download. There is also a full account export in JSON covering devices, alerts and settings. If you stop paying, the history you already have stays exportable.
$4.99 per device per month, or $49 per device per year. Every feature is included at both tiers — there is no plan where the useful tools are held back. The hardware is a separate one-time purchase, and you can cancel any month from the billing portal.
Put one in the ground
Stake it in the spot that browns first. Inside a day you will know how fast your soil actually dries; by the weekend you will know exactly how many minutes it needs. That is the whole product.
Not sure it fits your yard? Send us the layout and we will tell you honestly how many sensors it needs — including if the answer is one.