From Field Description to Printed Spray Card

The spray card is the piece of paper a spray operator fills out before every application. It lists the crop, the chemicals, the rates, and the label citations that prove the application is compliant. Filling one out manually means opening the product label, cross-referencing the crop, the pest, and the application rate, then transcribing all of it onto the form. For a rep doing multiple jobs in a day, that's 15-20 minutes per application.

We built a different path into the Freedom Ag portal.

A rep describes the job out loud: the crop, the pest pressure they're seeing, the products on hand. The AI assistant hears the description, pulls the relevant product records from the database, checks the label for the crop and pest combination, and produces a completed spray card: products, rates, PHIs, REIs, and the specific label citation for each item.

The rep prints it or sends it to the operator.

The label citation is the hard part.

A product label is not a short document. It's a 20-40 page legal document listing every approved crop, every approved pest, every rate range, every restricted-use interval, and specific language required on application records. "Follow label directions" is the legal standard. A spray card that can't cite the label is a compliance gap.

Pulling the right citation isn't a lookup. It's reasoning: this product, this crop, this pest, this rate. Is the crop listed? Is the pest listed under that crop? What does the label say the application record must include? The AI reads the label text, finds the section that applies, and writes the citation into the card.

Why this is not a chatbot.

A chatbot answers questions. Ask it "can I spray this product on soybeans for aphids?" and it tells you yes or no, maybe with a rate.

The spray card workflow doesn't answer a question. It reads the product database, reasons over the label, and produces a document the operator can sign and keep on file. Those are three separate operations on real data. The AI isn't summarizing: it's working.

This distinction is what makes building at the data layer matter. The assistant doesn't have product knowledge baked into its training. It reads the actual label records at runtime, which means when a label changes, the card reflects the current version. When a product gets added, it's immediately available. The assistant works from the live database, not a static snapshot.

The outcome.

A spray card that used to take 15-20 minutes to fill out now takes the time it takes to describe the job, which is under two minutes. More important: it cites the label correctly every time, which is the one thing manual entry consistently gets wrong under time pressure.

The Freedom Ag system handles this as a task the rep initiates through conversation and the assistant completes without further input. That's the pattern across every automation we build: the rep handles the judgment call, the system handles the known repeatable work, and the compliance gap closes.