Your watering decisions, handled.
Autopilot is your AI irrigation planner. Built on plant knowledge and models trained across multiple sites, then refined with feedback from your yard, it decides for every zone whether to water, when, and for how long — so you don't have to become an irrigation expert.
A good plan starts before the weather forecast.
Different plants need different care, and the same watering time delivers different amounts through different sprinklers or drip. Weather adjustments help — but weather alone can't tell whether the starting plan fits your plants and your watering system.
Seven inputs. One plan for each zone.
Autopilot combines your setup, local evidence and the forecast to decide whether to water, when to schedule it and how long to run. Your allowed times and limits remain part of the plan.
The drying pattern is learned from moisture readings and watering history — not from an extra sensor. A moisture trigger is one part of planning, not an unconditional command to open a valve.
Same water need. Different minutes.
Autopilot turns each zone's estimated water need into minutes using your sprinkler or drip parameters — then checks how the soil actually responded, and refines the next runtime.
One session can include time to soak
Autopilot schedules at most one planned watering session per zone per day. A session can include shorter watering cycles with soak breaks between them, so water has time to move into the soil before watering continues — useful on clay and slopes.
See how the plan responds.
Rain is only one input. Through a dry California or Arizona summer, plant settings, soil readings and drying patterns still guide each zone's next watering — so two zones under the same sun get different plans.
The rain can stop long before your soil needs more water. Soil readings, the zone's drying pattern and plant settings estimate when watering may be needed again, and new readings and weather keep that estimate current — not a return to the old schedule.
Start with your plants. We'll help with the settings.
Choose the starting point that works for you. Recommendations set up the zone; local readings and watering feedback help Autopilot plan what happens next. Where you can, group plants with similar needs on the same valve.
Low-water doesn't mean no-water — but giving a low-water planting the lawn's routine can add water it does not need. Start with the plant, then account for the site, not the habit of watering every day.
It starts with knowledge. It learns from your yard.
Plant knowledge and models trained across multiple sites provide the foundation. Your zone adds its own evidence — how watering changes the reading and how the soil dries afterward — and every new reading informs the next plan.
How the drying model works
After each watering, Autopilot studies how the moisture reading falls, combines suitable event curves, and uses the current reading to estimate when the zone may reach its watering trigger. The forecast and your rules help turn that estimate into a scheduled task.
Each drying curve is modeled as m(t) = m₀ − k · ln(1 + t/τ). The main planning path reviews irrigation events from the past 20 days. It combines predictions from qualifying event curves, weighting them by their above-trigger observation spans. The latest candidate is temporarily excluded if fewer than 40 hours separate its moisture peak and the latest available reading. A simpler linear calculation is available when suitable curves are unavailable or recent event performance does not meet the model's rules.
The estimated time to a trigger assumes no new rain or irrigation; new readings, weather and scheduling constraints can change the plan. 30% is the trigger used in this example, not a refill target or a recommended setting for every plant and soil.
Matching water to the plant and the soil is what saves it.
You set the boundaries. Autopilot handles the planning.
Keep control of allowed watering times and runtime limits. Pause, skip or run a zone manually when you need to. Autopilot plans within the settings you choose.