A vendor pitched you artificial intelligence. Before the cheque clears, run the pitch through eight questions. You’ll get a plain verdict — green-light, pilot carefully, or walk away — and the reasons behind it.
Most AI in agriculture works. It also fails — predictably. The difference is rarely the algorithm; it is whether the problem fits what these tools are actually good at.
The pattern, stated plainly: AI earns its keep on narrow, frequent, reversible decisions where you already have representative data. It disappoints on broad, rare, hard-to-reverse decisions in data-scarce settings — no matter how good the demo looked.
This filter turns that pattern into eight questions a manager can answer from a vendor conversation. It doesn’t score the technology. It scores the fit between the technology and your situation — which is the thing that actually predicts whether you’ll see a return.
A low score doesn’t mean the product is bad — it means the fit is poor for you, right now. The same tool can be a walk-away for one operation and a green-light for another. Re-run it whenever the pitch, the price, or your data situation changes. And remember the oldest rule in the book: if the vendor can’t tell you what decision their AI changes, there isn’t one worth paying for yet.
After the filter
The eight questions score the fit between a technology and your situation. They are answerable in four minutes from a vendor conversation, which is their value and also their limit: they use what the vendor told you.
A readiness memo runs the same test against your operation rather than against the pitch — your data, your decisions, your season, and what the system would have to change to pay for itself. It is a fixed-fee written opinion, addressed to you, that helps you decide whether to buy, pilot, renegotiate, or walk away. Where the answer is walk away, it says so and explains why. That is the point of paying for it.