Skip to content

[ ]

A team of AI agents runs your greenhouse.

Climate and irrigation agents run your crop on greenhouse physics, briefed early by weather intelligence. Every move is explained.

ghyll · agent consolelive
06:14forecastradiation ramp expected 09:40 · clear afternoon
06:52irrigation-agentfirst shot armed · radiation sum 112 J/cm²

How it works

Every agent runs the same loop

SENSE

Collect

Continuous real-time data from every sensor in your greenhouse. Always watching, always measuring.

DECIDE

Evaluate

A physics engine evaluates your greenhouse conditions and determines the optimal response. Deterministic and explainable.

ACT

Execute

Moves vents, valves, screens and more, always behind multiple safety layers.

LEARN

Adapt

Measures outcomes against predictions. What worked gets reinforced. Knowledge compounds across seasons.

↻ loops back

Sunlight, water and heat move fast. Ghyll's agents move first — every decision physics-checked, every action explained.

The agents

Meet the Climate Agent and the Irrigation Agent

[ ]

Handles heat before your crop feels it

Ventilation, shading, fogging and heating, steered continuously by the energy balance in your greenhouse, not by yesterday's setpoints.

  • Energy-balance-driven ventilation and shading
  • Disease risk scored live for four pathogens
  • Proactive condensation prevention
  • Night-to-day transition management
Zone climatelive
1.12kPa VPD
held in the crop's comfort band since 09:10
Leaf Temperature28.1°C
Humidity71%
Heat Load485 W/m²

[ ]

Water follows the sun

From the first radiation-triggered shot after sunrise to overnight dryback, the irrigation agent runs the full cycle, with volumes and timing matched to growth stage.

  • Radiation-sum trigger for the first shot of the day
  • Drain targets tracked shot by shot
  • Automated dryback management
  • A daily scorecard: first shot, sunset cutoff, drain and EC

[ ]

Sees the weather before it arrives

Ghyll reads tomorrow's radiation, temperature and rain hour by hour, then briefs the agents early. Shade screens deploy before the heat spike, and irrigation plans around tomorrow's light. Forecast accuracy is checked against reality every day.

  • Hour-by-hour radiation and temperature outlook
  • Shade screens staged before heat spikes arrive
  • Irrigation planned around tomorrow's light
  • Rain warnings and monsoon alerts built in
Radiation outlooknext 24 h
612W/m² peak
tomorrow 12:40 · screens staged ahead
nowscreen stagedlast shot
06
40
08
180
10
430
12
610
14
540
16
300
18
90
20
0

[ ]

The intelligence behind every agent

Ghyll Intelligence is the shared foundation the agents run on: greenhouse physics, science-backed crop models, and what your own greenhouse has taught the system, season after season. If you use Ghyll, you're already using it.

Ask Ghyll is the direct line to it. Ask about today's decisions, your crop's trend, or next week's plan, and get answers grounded in your greenhouse's own data, never generic advice.

  • Physics and crop models, not black-box rules
  • Learns your greenhouse and remembers next season
  • Every answer cites the data behind it

[ ]

grounded in your greenhouse data
Why did irrigation start later this morning?
Cloud cover kept the radiation sum below the 110 J/cm² trigger until 08:15. The irrigation agent fired the first shot at 08:22, about 90 minutes later than yesterday. The morning weather briefing had already planned for it. Substrate moisture stayed inside its target band all morning.
sources: irrigation-agent log · weather briefing 06:14 · zone 2 sensors
Ask anything about your crop…

Why Ghyll

Grow more with less. Consistently.

Higher yield

Closed-loop autonomy

Decisions follow the crop's physics, not a timer. The system responds to what the plant experiences, every minute of every cycle.

Lower inputs

Less water, energy, fertilizer

Water, energy and fertilizer are spent only when the crop needs them. Waste is engineered out, not trimmed after the fact.

Every decision explained

Autonomy you can audit

Each action carries its reasoning: what the agent saw, why it acted, what happened next.

Multi-crop

One platform, any crop

Start with science-backed crop profiles, then configure setpoints for your greenhouse. The agents carry the rest.

Light needs

Water sensitivity

Nutrient intensity

01. Blueberry

Southern Highbush. A pre-built profile with science-backed setpoints, ready to deploy.

Agentic AI that earns your trust

Physics makes the decisions from day one. Self-improving intelligence learns what works in your greenhouse and remembers it next season. Every decision explainable. Every outcome measured.

Ready to see it in action?

Schedule a demo and watch Ghyll's agents run a greenhouse, then ask them why they did it.

Request a Demo