What intervention maximises expected crop value?
Not the largest biological effect — the largest economic one, net of what it costs to produce.
From AI assistant to agricultural agent
Agricultural autonomy does not require removing the grower.
It requires giving the grower a fundamentally more powerful operating layer. The destination is not an AI that tells a grower what happened — it is an agricultural intelligence capable of helping determine what should happen next, inside boundaries a human sets and can revoke.
Instrument 06 — The autonomy ladder
Monitoring → Diagnosis → Prediction → Optimization → Autonomy
The progression is deliberate. Select a level to see exactly what moves to the machine, what stays with the operator, and which governance modes even apply — because below Level 7 the system holds no authority to act at all.
Observe
What the machine does
What stays with the grower
Governance mode
Operating authority
4% machine · 96% operator
System output at this level
Select a level, or focus the rail and use ↑ ↓. Console output is illustrative.
Human-governed autonomy
Autonomy is granted, bounded and revocable — never assumed.
The grower moves from manually interpreting every signal toward supervising an intelligent cultivation system. Human expertise becomes amplified by machine-scale perception, memory and analysis.
The grower supervises. The system executes only what it has been permitted to.
01Recommend only
AI identifies conditions and recommends action. Every execution decision stays with the operator. The system’s value here is entirely in what it notices and explains.
02Approval required
AI proposes an intervention and an authorized operator approves execution. Nothing reaches a physical system without a named human authorization, recorded in the audit trail.
03Bounded autonomy
AI may independently adjust predefined systems within operator-defined limits. The envelope is explicit — approaching its edge escalates to a human rather than expanding it.
04Autonomous optimization
Validated environments can permit increasingly autonomous operation while preserving human oversight, safety constraints, auditability and manual override at all times.
Configurable autonomy means the operator decides how much authority the system has — and can take it back in a single action.
Instrument 07 — Closed-loop cultivation intelligence
The step most systems skip is the one that matters: verify.
A recommendation engine ends at the recommendation. A closed loop observes what the crop actually did next, writes the outcome to memory, and lets the result change how the next decision is made.
The ring follows the live cycle. Select any stage to hold the readout.
The result is a continuously evolving feedback loop between biology, environment, machine intelligence and physical infrastructure.
The economic intelligence layer
Agricultural optimisation ultimately has to produce economic outcomes.
The platform should therefore evolve beyond biological optimisation alone. The future Master Grower is not simply an agronomist encoded in software — it is an agricultural operating intelligence reasoning simultaneously about biology, resources, operations and economics.
Which environmental adjustment justifies its energy cost?
A 0.4% yield gain for an 11% energy increase is a decision, not an optimisation.
Where is labour being deployed inefficiently?
Scouting hours spent walking rows that were already nominal on the last three passes.
Which zones are underperforming?
And whether the cause is environmental, operational, structural or genetic.
Which crop characteristics correlate with higher yield or quality?
Learned from this operation’s own outcomes, not from generalised agricultural literature.
Which interventions repeatedly fail to generate economic benefit?
Standard practice that has quietly stopped paying for itself.
What is the financial consequence of delaying intervention?
The cost of waiting, quantified — which is what turns an alert into a decision.
How should resources be allocated across a portfolio?
Across multiple facilities, where the same capital produces different returns.
Next
Cannabis is the proving ground. Agriculture is the market.
Where this intelligence is being tested first, and why that choice makes the platform stronger rather than narrower.