Hacker News Reader

Comment by lukeqsee | Hacker News Reader

I work at a precision ag startup in the Midwest. I can speak from inside the industry: it's an incredibly exciting time.

We have more people asking us to build tools than we could ever build, and most of the tools are basic management level. We haven't even scratched the surface of applying algorithms to the loads of data being created. I think over the next few years data + machine learning will make things like pesticide application, irrigation, fertilizer, and planting incredibly more efficient. The cool part is the growers have the tools to apply these algorithms as they're produced (tractors, combines, planters are almost all able to support variable rate applications). All we need is the ability to capture, analyze, and start applying the data.

Feel free to ask if you're more interested in what the industry looks like from the inside.

View on HN

Another Ag founder here and this is totally true. There's a lot of groundwork to lay before we can start applying state of the art algorithms to ag.

For example, a few months ago my co-founders and I built a computer vision system that could count fruit (predicting yield is important because produce spoils quickly). The growers we've shared it with love it but most don't have the infrastructure to use the data in their operations. It's hard to respond quickly when everyone is emailing spreadsheets.

Adoption of tech in ag has been slow, in part, because farmers are already very good at their jobs. Many of them have been working the same crop on the same land for decades so there has been some skepticism about what new information can be learned from "just another sensor."

But change is in the air. The water is running out, there's a huge labor shortage, and every farm employee now has a smartphone.

If we can spend the next year or two getting farm operational data online then there is a huge amount of value that can be unlocked with even relatively simple algorithms (not to mention deep learning etc.)

In fact, we don't really have any other choice. Software can bring in the next green revolution and that makes this a really exciting time.

Absolutely agree. Everything you just said I've heard or experienced in just a few months.

> There's a lot of groundwork to lay before we can start applying state of the art algorithms to ag.

We have experts in algorithms, embedded systems engineering, computer vision, software engineering, sensor technology, and robotics at our company (along with a few other specializations). But before we can utilize that talent to its fullest, we have to build the foundation. So that's what we, and a whole bunch of other really talented companies, are doing: building the foundation. Hopefully in a few growing seasons we can start to really apply what we've learned.

Are you open-sourcing your systems so that others may benefit? Or keeping them locked down to extract maximum profit?

I downvoted you because this strikes me as gratuitously negative, mean-spirited, and/or uncalled-for. Asking for pointers to any open-source work they've done would have been more appropriate.

What exactly is negative in my comment? Is extracting max profit bad? Or open-sourcing work?

The way you worded it sounds very negative, as if the only reason for keeping thing locked down is to extract maximum profit.

If you had said something like "fund the work" it would have sounded more acceptable. There is a large space between "extracting maximum profit" and "doing work for free".

What other reason is there for keeping something locked down?

Off the top of my head, scientific publishing. Keep it locked down until you have published then open source it.

I personally wouldn't want my work code online, as its always rushed and there is a fair bit of technical debt I would like to to refactor, rather than having potential employers thinking that I always write shit code. Feature requests are valued more than code quality.

I suppose wanting to hide perceived 'shoddy' workmanship is a reason.

I say, let it all hang out. Code can be continually improved.

I happen to think that max profit is a worthy goal. And big ups to any one that takes that route. It's that sort of thinking that drives efficiencies that make all our lives better. So when I read that comment I don't view it as negative. That said I see how it could be taken as negative.

You describe those options as though they are at opposite ends of the same spectrum. I'd say they're somewhat orthogonal.

> We have more people asking us to build tools than we could ever build

What are a few examples?

Our current product was just released to address the needs of one market (scouting companies, if you aren't familiar with the term: http://www.farms.com/precision-agriculture/crop-scouting/). We are also building a general-purpose platform for handling the data (and partial visualization) of the farmer, e.g., field geolocation data, various data layers, etc.

The first addresses one market. The second addresses the entire market at a high-level. Most individual markets (such as seeds, fertilizer, pesticide, etc.) need specialized solutions.

For instance, fertilizer companies and growers would love a tool that lets them take in all of their data and using known models (which exist in theory in academia) apply that information to allow them to apply correctly across a field. A myriad of individualized solutions for locales, crops, and the ilk are just begging to be built. And that's just fertilizer, one of the many cycles in crop production.

Large and small corporations are asking us to do work on various pieces of that huge pie, but we have to specialize or our fairly small team would be impossibly overloaded.

>>All we need is the ability to capture, analyze, and start applying the data.

Who owns the data?

That is a worthy and unanswered question in the industry. I'm personally hoping that the data becomes free and open (OADA is an attempt to support that); I believe, along with at least a few of my colleagues, that the "secret sauce" is in the algorithms and tools to utilize the data.

We shall see.

I would love to hear about what you guys are doing. I think thats a huge underrepresented field (outside of monsanto/cargill).

We are specializing on one stage of the crop-growing cycle: pest management.

In order to do that, we started by building a hardware tool to deploy in fields to monitor moth population. That's how incredibly specialized this industry requires: one part of the cycle and for one pest merits an entire company being started.

Because hardware is really hard to get right, and then to scale, we've started developing software products in parallel. What that means is like just about every other provider in the space, we had to develop a general purpose platform to allow growers to even see their data (farm, field level management and visualization). Luckily, we have a contract with a regional company to build that tool while maintaining the IP. We're just now starting to get that right, so that we can move on to solving interesting problems in the space (we just released our first product built on the platform for scouting companies, see: https://news.ycombinator.com/item?id=9331917).

Thankfully the data problem is in the slow progress of being solved thanks to http://openag.io/. Hopefully in the future, new companies only need to develop awesome solutions to particular problems and allow those to be integrated with a variety of data storehouses. Unfortunately, that's a long way off.

do you work at farmlogs? its the only one I've heard of.

I don't. But if things go as planned, you'll be hearing about us soon. :)

Reply on news.ycombinator.com