Farmers leave 20–40% of their profit on the table by overwatering — not from carelessness, but because every irrigation call carries uncertainty, and uncertainty compounds into systematic waste.
Kahneman calls it noise. The fix isn't more data — it's a decision system that turns data into one number (that informs your gut-feeling - NEVER IGNORE THAT).

For operators & investors:

The rest of this piece is how I got there — through a small Australian software company, a sponge, and a chart that explains why "better safe than sorry" is the most expensive phrase in agriculture (and it's probably in your business or investments too).

Most people think drought is farming's biggest enemy.

It's not.

Here's the paradox: most farmers leave 40% of their yield on the table — because they use too much water.

And the water they're wasting?
It's running out.

Three years ago, Google's AI grew better tomatoes than specialized hydroponic software. This gives a glance into the future of watering.

Here's why farmers overwater, what it costs them, and what it tells us about every decision we make under uncertainty (investing included).

We call Earth the blue planet.
It looks blue from space. It's 71% water.

But 2.2 billion people still lack access to safely managed drinking water.

We can only use 0.5% of the water on Earth.
70% of that goes to agriculture.
And only 40% of that actually reaches the plant.

Which means humans waste 42% of all freshwater — mostly into the ground.

By 2030, demand is projected to exceed supply by 40%.

Microsoft and PepsiCo are already investing in agricultural water efficiency.
They see what's coming.

Given all this, you'd expect farmers to guard every drop.

They don't.

Imagine you're a farmer.
Water is maybe 5% of your total costs.
A bit extra looks like a rounding error.

But the farmers who get irrigation right use 10–30% less water than you — and see 20–40% higher profitability per hectare.

How does 5% of costs unlock 40% of profit?

Because overwatering doesn't just waste water.
It actively destroys yield.

Plants are like us: they need water and oxygen.
Too much water, and they can't breathe.

As a farmer, you know this.
You do it anyway.

You know the soil is a sponge.

Pour water on top.
The sponge absorbs it.
The plant drinks.

Keep pouring, and water drains out the bottom.

Same as us humans: too much water, you lose nutrients.

Except for a plant, the fertilizer is dissolved in that water.
Keep pouring, and you flush it past the roots — into the groundwater.

Lose-lose: wasted inputs, polluted drinking water, smaller harvest.

Tim Hyde, an Australian grape and banana farmer, gave me the analogy I can't unsee:

Countries have researched optimal irrigation for 20–30 years.
This isn't new science.

So why do farmers keep overwatering?

Here's the situation every planting season:

You have a crop worth $20,000 per hectare.
A heat wave is forecast.
Your data probably says the soil is humid enough.

Probably.

You're not really sure.
So you irrigate.
Just to be safe.

Four or five times a year — 10 to 30% overwatered.

The problem isn't data.
We've had soil moisture probes since the 1970s.
The problem is interpretation.

Optimal irrigation depends on soil type, root depth, variety, growth stage, weather, actual water applied — all synthesized into one number.

Sensors give you data.
They don't tell you what to do with it.

That little insecurity — am I reading this right? — makes you add a bit more.
Your personal insurance policy.

And this is where it gets interesting.

Daniel Kahneman (Author of Thinking, Fast and Slow, and later Noise) has a specific word for this.

Every farmer has an uncertainty range — a zone where they think the right answer probably lies.
Ask ten farmers the same irrigation question and you'll get ten different answers.
That spread is noise.

The average amount by which farmers systematically over- or underwater is bias.

Kahneman's key finding: reducing noise has the same effect as reducing bias.
You don't need to be smarter.
You just need to be more consistent.

Here's what it looks like on a yield curve:

Optimal profit sits right around 100% of required water.
A farmer with a tight decision process (Farmer 2) lands at ~120% and captures 44% profit.
A farmer with a wider uncertainty range (Farmer 1) pushes to 150% — "better safe than sorry" — and walks away with 35%.

Same crop.
Same field.
Same weather.
Nine percentage points of profit, gone — purely to noise in the decision process.

Now remember Google's tomatoes.
The reason their AI beat specialized hydroponic software wasn't a smarter model.
It was a more consistent decision process.
Same insight.

So the question becomes: how do you strip noise out of an irrigation decision?

Kahneman's answer is simple: build a decision system.
A repeatable process that turns data into a single number.

Which is exactly what a small Australian software company figured out.

Two reasons.

First: most ag-tech sells hardware.
Farmers don't want to rip out their existing probes, flow meters, and controllers.

Second: the industry is fragmented.
Soil probe on one platform.
Flow meter on another.
Weather on a third.
Each vendor wants you on their dashboard.

Nobody aggregated it into a single answer.

Tim Hyde felt this personally as a grape and banana farmer.
He and a few others went out to fix it.
That became SWAN Systems.

Here's what they did differently:

Not "here's your data." But "here's your decision."

A co-pilot for irrigation — the decision system Kahneman would have prescribed.

In July 2024, WSU ran a peer-reviewed study on an apple orchard using SWAN.

The result:

That doesn't read like a tradeoff.
That reads like "we were leaving this on the table for decades."

Real customer data tells the same story:

So why aren't they a billion-dollar company already?
Let's look at the financials.

Solid metrics for a SaaS that's only been commercial for five years.

The risks:

B2B sales cycles in agriculture are long — 1 to 2 years for enterprise deals.
SWAN is still pre-profit, targeting $3–5M in revenue.

30–40% of revenue comes from wine grapes, an industry under pressure.

Risk acknowledged.
But when margins compress, efficiency software stops being a nice-to-have and becomes survival software.
That's when SWAN crosses from optional to essential.

Today, SWAN tells farmers what to do.
They still have to program the controller themselves.

Their upcoming feature, Swan Sync, removes that step.
It generates the optimal schedule and pushes it directly to the controller.

Two clicks.
Done.
First-in-world functionality.

The only open question: will farmers trust AI to make the call for them?

There's a saying in the industry: AgTech scales at the speed of trust.

SWAN spent five years earning it.
Now they're converting it into automation.

Strip away the irrigation and here's what's left:

A recurring decision.
Under uncertainty.
With real money on the line.

Sound familiar?
That's every investment decision you've ever made.
Every hiring call.
Every pricing choice.
Every product you launched.

Kahneman's point — and the whole reason this video took me down a rabbit hole — is that the biggest gains don't come from being smarter.
They come from being more consistent.
From turning a judgment call into a repeatable process.
(--> making sure, if you got the same data, your conclusion doesn't differ widely. It often does even if we don't think so)

Whenever you have a decision you make over and over, build a system for it.
If this always changes, build a comparison table for the inputs.

That's the whole lesson.
That's what SWAN sells to farmers.
That's what a good investment checklist does for a fund manager.
That's what a written trading plan does for a solo investor.

The moat often isn't the data.
It's the decision system sitting on top of it.

Build the system first.
Then automate.

🎥 Watch the full video featuring Tim Hyde, CEO of SWAN Systems: https://youtu.be/Rzq9jdhNANE

Disclosure: I'm compensated for this video in connection with the MK Investment Conference — not by SWAN Systems directly. You should know that before forming your own view.

Know someone who'd find this interesting? Forward this email — they can subscribe for free.