Agriculture Software Development

CodigoDelSur builds the software that agriculture companies put in front of growers, agronomists and field crews: drone flight and image capture apps, crop scouting and pest management tools, dairy operations platforms, and the trading and traceability systems that run after the crop leaves the field. We have been building software from Montevideo, Uruguay since 2007, for companies in the United States and the rest of the world.
This page is not a list of industries we say we serve. It names six agriculture products we have worked on, says what our engineers built and what the client built, and says how long each engagement ran. It is not a full inventory of our agriculture work: it is the part we can show in public, which is the only part worth reading, and it is the part you can check without taking our word for it.

Six agriculture products we have worked on

The most useful question you can ask an agriculture development partner is which part of the product was actually theirs. Detection models, agronomic thresholds and flight hardware often belong to the client, and finding that out after you have signed is expensive. Here are six of ours, with the boundary stated in every row.

ClientThe productWhat our engineers builtStack
Milc GroupDairy operations platform with modules for feed, facility, animal and people managementiOS, Android, web and backend across five business domains, integrations with the dairy systems the client already ran, and MIRA, the computer vision assistant, including the detection models and the labeling and retraining pipelineSwift, SwiftUI, Java, Node.js, React, Python, AWS, YOLO, Roboflow
BluefinAgriSite IPM, integrated pest management that moves scouting observations between scouts, agronomists and producersThe iOS application and the field workflow that connects the three rolesObjective-C, Swift, C#, Xamarin
AgroScoutPrecision agriculture platform that detects pests, disease and crop stress from drone, satellite and phone imageryThe mobile application, including a custom flight control loop on the DJI SDK that holds altitude over sloped terrain and triggers the camera at fixed intervals. The pest and disease detection models are the client's own work, not ours.Objective-C, Swift, DJI Mobile SDK, Mapbox, Android
Skippy ScoutDrone crop scouting: automated flights that photograph fixed scouting points across farms and fieldsThe iOS application, including the automated flight and image capture behaviourObjective-C, DJI SDK
DroneDeployDrone mapping and reality capture, with automated flights, pre-flight and in-flight safety checks and flight logs used for compliance, insurance and fleet maintenance, across agriculture and constructionThe native iOS and Android flight applications on the DJI SDKSwift, Objective-C, DJI SDK
Caravela CoffeeAtlas, the internal platform for coffee buying, export, import and lab testing, and Roaster Hub, where roasters source green coffee and manage contracts, inventory and procurementBoth platforms, web and backendRuby on Rails, React, PHP, Azure, Sidekiq, Twilio

Where the line falls on AI, in the two projects where it matters

Two of these six are known for their machine learning and the answers are opposite, which is exactly why we publish both. On AgroScout the pest and disease detection is the client's own work and we built the mobile application around it. On Milc Group we did the computer vision work ourselves: our engineers trained the detection models with YOLO and built the labeling and retraining pipeline in Roboflow. The models belong to Milc Group, as everything we build does. Any partner can put a client's AI on their own capability slide. Ask which one you are being shown, in writing, before you sign.

Bluefin is the one client on this list without a case study page on this site. Rather than link you to something generic, we would rather say so.

Five things agriculture software has to handle that most software does not

Not a list of trends. Each of the five is a constraint we hit on one of the projects named above, and it is the reason that code looks the way it does.
The ground is not flat. A drone that photographs a field has to hold its height above ground while the ground itself rises and falls. DJI's standard waypoint missions cannot adapt to elevation in real time, so on AgroScout we bypassed mission planning entirely and took direct, low latency control of the aircraft, reading telemetry every 10 milliseconds and recalculating the flight as it went. Terrain is a software problem before it is an agronomy problem.
The image is the measurement. In scouting software the photograph is not documentation, it is the data the analysis runs on, so spacing and interval matter more than resolution. AgroScout triggers the camera every 15 seconds at a held altitude; Skippy Scout automates flights to photograph fixed scouting points across a field. Gaps between waypoints show up months later as gaps in the agronomic record.
You are building on somebody else's SDK. Three of the six products above run on the DJI Mobile SDK, which means part of your release calendar belongs to a hardware manufacturer, and firmware and SDK changes reach your users before they reach your roadmap. The useful question to ask a partner is how many drone SDK versions their code has survived, not whether they have worked with drones.
Nobody in the field is the same user. In pest management a scout records what is in front of them, an agronomist decides what it means and a producer decides what to do about it, and the three are rarely in the same place or on the same schedule. Bluefin is that workflow, built as one product for three roles. Software designed around a single user does not survive contact with a farm.
A lot of agriculture software never touches a field. Buying, export, import, lab testing, contracts and inventory are agriculture too, and they look like enterprise software rather than like a drone app. Caravela Coffee's Atlas and Roaster Hub are that side of it, and Milc Group is the same problem inside a dairy: modules that have to agree with each other and integrate with the systems the client already runs.

How long these engagements actually ran

Agriculture runs on an annual cycle, so a partner who leaves after one season never saw the product work. Engagement length is the number agriculture buyers ask for least and should ask for most. These are the agriculture engagements on this page.

ClientHow long
BluefinTwelve years and counting, since November 2013
Milc GroupNine years, with a dedicated team that grew to 40 people across five business domains
AgroScoutThree years and nine months, January 2021 to October 2024
Caravela CoffeeThree years and eight months, March 2022 to November 2025
DroneDeployTwo years, 2016 to 2018
Skippy ScoutOne year and four months, September 2019 to January 2021

Milc Group is the only row without an exact range: the engagement ran nine years and we do not hold the start month in a form we would publish, so we say nine years rather than invent a date. A partner who rounds these to a convincing number is telling you something about the rest of their numbers.

The longest relationship in the company is not in agriculture and is not on this page: the first client CodigoDelSur ever signed, in 2007, is still a client today.

What we build in agriculture

Five categories and not fifteen, each one named with the product it shipped in. If something you need is not on this list, it is not because we forgot it.

What we buildWhere it shipped
Drone flight and image capture applications, including custom flight control on the DJI SDKAgroScout, Skippy Scout, DroneDeploy
Field scouting and pest management tools built for more than one roleBluefin, AgroScout
Farm and dairy operations platforms, including integrations with the systems the client already runsMilc Group
Computer vision on field and animal imagery, including the labeling and retraining pipelineMilc Group
Trading, traceability and supply chain platforms for what happens after the harvestCaravela Coffee

What working with us on an agriculture product looks like

How fast a team starts. Five business days from a signed agreement to the engineer's first day is the standard, and as soon as the next business day if the profile is already on our bench and your side is ready. The minimum term is two weeks, one sprint, and the minimum project size is $10,000, roughly one engineer for one month.

Hours, and why they matter more here. Our engineers work 10:00 to 19:00 Uruguay time, which covers the entire US East Coast working day and five to six hours of live overlap with the West Coast. Field software raises the stakes on that: a flight bug reproduces at the field, in daylight, with a drone in the air, and a partner who is asleep at that hour costs you a full day per iteration rather than an hour.

If an engineer is not working out. You tell your account manager and we replace the person at our cost. The outgoing and incoming engineers overlap for ten business days and we do not bill those days, so context and work in progress transfer between them instead of landing on your team.

What happens at the end. Everything we build is yours, with full intellectual property assignment, including any model we train on your data. The Milc Group engagement is the one to ask us about: nine years, a team of 40 at its peak, and an ordered handover to the client's own engineers when they were ready to run it themselves. Whether a partner has ever exited an engagement cleanly is a better question than whether they promise to.

Three ways to engage. Agriculture work is delivered through the same three models as everything else: staff augmentation, a dedicated team, or full-cycle product development. The table below is the difference between them.

If you are earlier in the process than this page assumes and you are still building a shortlist, we published a comparison of five agtech development partners, ourselves included, with the selection criteria stated and with the ones we lose on named.

Agriculture is one of the industries where we can show the work in public. The rest are on the industries page, along with the five questions worth asking any development partner before you sign one.

If the part you need help with is the flight rather than the field, the drone software page covers the aircraft side: flight control on the DJI SDK, airspace and hardware validation, and the flight logs your customers will need later.

Staff augmentation, dedicated team or full-cycle product development

Three different products get sold under the same word, "outsourcing". They are not interchangeable and picking the wrong one is the most expensive mistake in this category. Agriculture work can be delivered through any of the three, and all six products above were.

Staff augmentationDedicated teamFull-cycle product development
Who owns the roadmapYouYouShared, we run discovery
Who manages day to dayYour engineering managerYou, with a CodigoDelSur project manager coordinating on our side at no extra costOur project manager
What you getIndividual engineers inside your teamA team shaped around your product, staffed and coordinated by usA product, from discovery to store release
You need to haveTechnical leadership with spare capacityA product owner and a clear roadmapA budget and a problem worth solving
Who owns the field data, the imagery and any model trained on itYou, alwaysYou, alwaysYou, always
Typical size1 to 4 people4 to 12 peopleScoped per project
Minimum commitmentTwo weeks, one sprintTwo weeks, one sprintQuoted per project
Best whenThe work is defined and you are short on handsYou own a product area and want it staffed and coordinated as one teamYou need the thing to exist and you do not have an engineering org

When this page is not your answer

If what you need is agronomy rather than software, we are the wrong supplier and it is cheaper to find that out here. We do not produce agronomic models, spray thresholds or crop science, and we do not employ agronomists. What we do is build the application around yours, and AgroScout is the clean example of that division: the detection models are the client's own work, and we built the flight software around them. If nobody on your side owns the agronomy, hire that before you hire us.

Where "taking over an existing project" and "short-term work" fit

Neither of those is a fourth model. They are situations, handled inside the models above. If you already have a field application built by someone else and want to switch providers, we review the existing codebase and work on top of it, either with engineers embedded in your team or with a dedicated team taking the product over. If the work is short, the two week minimum means a single sprint can be the whole engagement.

Frequently asked questions

Do you build the agronomy, or the software around it?

The software, in almost every case, and we say which is which project by project. On AgroScout the pest and disease detection models are the client's own work and we built the mobile application and the flight control around them. On Milc Group we did the computer vision work ourselves: our engineers trained the detection models with YOLO and built the labeling and retraining pipeline in Roboflow, and the models belong to Milc Group, as everything we build does. We do not employ agronomists and we do not produce crop science. If a partner will not draw that line for you in writing, assume it falls in whichever place flatters them.

Have your engineers actually shipped drone software, or just mobile apps?

Three products on the DJI Mobile SDK: AgroScout, Skippy Scout and DroneDeploy, the last of which is used on more than 3 million jobsites. On AgroScout the standard waypoint mission mode was not enough, because it cannot adapt to elevation changes in real time, so we bypassed mission planning and took direct low latency control of the aircraft, reading telemetry every 10 milliseconds and triggering the camera at a fixed interval while hugging the terrain. That is the level of the work, and it is the part that is hard to fake in an interview.

What kinds of agriculture software do you build most often?

Four things, in this order. Drone flight and image capture applications. Field scouting and pest management tools that connect scouts, agronomists and producers as separate roles in one workflow. Farm and dairy operations platforms, including integrations with whatever the client already runs. And trading, traceability and supply chain systems for the part of agriculture that happens after the harvest, which looks like enterprise software rather than like a field app.

What happens when the engagement ends?

You own everything, and we hand it over. The clearest example is Milc Group: nine years, a dedicated team that grew to 40 people across five business domains, ended with an ordered handover to the client's in-house engineers when they were ready to run it themselves. Our agriculture engagements have run from three years to twelve and counting, and the shape of the ending is worth asking every partner about, because it is the only part of the relationship they have no incentive to plan.

What time zone do your engineers work in, and how much overlap will I get?

Our engineers work 10:00 to 19:00 Uruguay time, which is UTC minus 3. That covers the entire US East Coast working day, five to six hours of live overlap with the West Coast, and three to four hours with Europe, in their afternoon. The schedule is set that way deliberately: shifting the day an hour later widens the Pacific overlap without asking anyone to work nights, and it is the reason this arrangement is still working in month twelve rather than quietly falling apart in month three.

How fast can you actually start?

Five business days from signed agreement to first day is the standard, and if the engineer is already on our bench and your side is ready to receive them, they can start as soon as the next business day. What varies is finding the right person, not starting them: if the profile is on our bench you get CVs one business day after you sign off the role brief, and if the role has to be recruited, first CVs take about five business days. We tell you which of the two you are in on the first call, before you have signed anything, because that is what actually moves your start date.

What happens if the engineer is not working out?

You tell your account manager and we replace the person at our cost. There is no charge for the search and no penalty for asking. The outgoing and incoming engineers overlap for ten business days so that context, access and work in progress transfer inside the engagement instead of landing back on your team, and those ten days are not billed: during the handover you keep paying exactly what you were paying for one engineer. We would rather absorb that than have you carry a bad fit for a quarter, and a company that will not put a replacement policy in writing is telling you something.

Who owns the code and the intellectual property?

You do, exclusively. Everything created for you as part of the work is assigned to you with full title guarantee: source code, application logic, APIs, integrations, data models, configurations, UI and UX designs, documentation and any derivative works. The assignment covers the deliverables as they are created, it is not conditional on final payment, and it is not a licence back to you, it is ownership. We also warrant that what we deliver does not infringe any third party's intellectual property, and where third party material is involved we secure the consents for both sides first. For our own pre existing components that were not written specifically for you, you get a non exclusive, irrevocable, royalty free licence to use, copy and modify them, so nothing we bring with us can be used to lock you in later. Every CodigoDelSur employee signs a non disclosure agreement and an intellectual property assignment as a condition of employment. In an agriculture engagement that also covers the field data you collect, the imagery it comes from and any model our engineers train on it, and we do not reuse one client's data or models on another client's project.

Tell us what you are building for the field

Tell us what the product has to do and where it runs: a drone, a phone in a field, a dairy office, a trading desk. On the first call we will tell you which parts we have built before and which parts we have not, and that second list is usually the more useful one. CodigoDelSur has been building software since 2007, from Montevideo, Uruguay, for companies in the United States and the rest of the world.