Our Work

Greenhouse Management for a Plant Farm

Our customer is a farm that grows and sells plants, and they needed a better way to keep track of them, from the day a seed goes in the tray to the day a plant goes home with a customer. Nothing off the shelf fit the way they work, so we're building it for them. The software belongs to them, and it's nearly ready for the farm to put to work.

The farm would rather keep their new system to themselves for now, so we've left their name off this page. Everything else here is shared with their permission.

At a glance

  • 115,000 lines of code, give or take
  • 680+ automated tests
  • 3 weeks of development time
  • 3 devices from one app: iPhone, iPad and Mac

Work started on October 20, 2026. About four weeks have gone by on the calendar since then, and at least one of those weeks went to the farm testing what we'd built so far and telling us what they thought. That's normal, and it's time well spent. So the development itself comes to roughly three weeks.

Why plants are hard to keep track of

Most inventory software assumes a thing sits on a shelf until somebody buys it. Plants don't work that way. They grow. You take cuttings from one and end up with a hundred more. Some of those cuttings don't root. The ones that do get potted up into bigger containers, moved to another greenhouse, priced, and sold a few at a time. A plant can change size, location and value several times before it ever leaves the farm.

So the software tracks plants in groups, the way a grower actually thinks about them, and every group remembers where it came from. Pick any plant on the sales bench and you can trace it back to the stock plant it was cut from, along with everything that happened to it in between.

What we built

The project has two halves that work together.

The server is where the farm's records live. It's written in Rust and uses a PostgreSQL database. It runs on a server right there at the farm, so their data stays in their own building, and staff can still reach it securely from anywhere without the farm's network being opened up to the internet.

The app runs on iPhone, iPad and Mac from a single codebase. It keeps working without a signal, which matters in a metal greenhouse, and catches up with the server once it's back in range.

Between the two, the system covers:

  • Inventory from start to finish. Plants that were bought, sown from seed, or propagated by cuttings, division, grafting and the like, through every stage, pot size and move until they're sold
  • More than one location. Separate farms or garden centers, each with their own greenhouses, benches and growing areas
  • Staff and customers. Owners, managers, employees and customers each see only what they should, and staff sign in with two-factor authentication
  • Sales and holds. A storefront where signed-in customers can see what's available and put plants on hold, plus a connection to Square so sales at the register come off the inventory automatically
  • Care guides. One set of care information for each plant, used everywhere it shows up. AI can write a first draft, and a person checks it before anyone else sees it
  • Labels. Plant tags and labels designed in the app and printed from the farm's own printers, with the plant's details and photo filled in
  • Reports. Losses, sales, how well cuttings are rooting, how long plants spend in each stage, and what the inventory is worth over time

We also brought over every plant from the farm's existing website, photos and all, so they didn't have to start from an empty system.

We're keeping the details of how it's put together to ourselves. That's the farm's business, and theirs to share if they want to.

How we went about it

A job this size goes wrong when somebody tries to build it all at once. We broke it into stages, around thirty of them across the server and the app. Before any stage started, we wrote down what it would do, and when it was done, it got tested and reviewed before we moved on to the next one. Some stages were small enough that the next step was obvious. Whenever we reached something the farm could really sink their teeth into, we stopped and handed it over so they could try it out for themselves. What they found shaped the stages that came after.

Every piece of it is covered by automated tests, more than 680 of them, and they run every time something changes. When the farm asks for a change six months from now, those tests are how we'll know the change didn't quietly break something else. We also went back over the whole thing, top to bottom, looking specifically for security problems, and fixed what turned up. When software holds a business's customers, sales and payments, that step isn't optional.

We used AI tools to help write the code, and we're not shy about saying so. That's a big part of how a project this size fits into a few weeks instead of several months. But the tools don't decide what to build, and they don't get the last word on whether it's right. Every stage was planned, checked and tested by us. We wrote more about how we think about that in Using AI Without Compromising Integrity. The short version is that the time saved goes to the customer, not into our pocket.

Where it stands

The planned work is nearly finished. The farm has been testing it along the way, and the next step is putting it to work every day in the greenhouse. Daily use always turns up things testing doesn't, and we expect to make some adjustments once it's part of the routine. We'll update this page when it's in full use.

What this means for you

You might not grow plants. But if your business has a way of working that the software on the market just doesn't fit, you're in the same spot this farm was. A system like this one, with a server, an app on several devices, payments, printing, reports and the security to go with it, is well within what we do. And it's all done here in the United States, by the same people you talk to.

Tell us how your business works and where your current software falls short.

Start the conversation