Grain is critical, and now it's sexy (again)

Blue Grain Fieldv2.png

The digital transformation of agriculture is not a nice-to-have, but a critical investment

For 10,000 years the world has been fed by grains; mainly wheat, corn, and rice. Only a few decades ago world population growth trends were leading people to predict massive starvation, but the Green Revolution of the 1960’s resulted in crop yields rising from 1.4 tons to over 3.2 tons per acre today. Even so, with wheat consumption forecast in 2018/19 to be higher than production for the first time in a decade, changing weather patterns, and a technology revolution rolling through the world of agriculture, today’s grain organizations are in a heightened state of digital transformation. 

We have just published a whitepaper that analyses the state of the market, to see who the winners and losers might be in the coming grain revolution. It’s free to download, but to whet your appetite, (while there’s still enough to eat) here are a few highlights:

  • China are the largest grain producing country in the world. It is the biggest producer of rice, but also produces more wheat than either America or Russia

  • Cargill, one of the largest global grain organisations began their 2018 Annual Report by saying ‘Why does a food and agriculture company start thinking like a technology startup?”. The simple answer is because they have to, in order to stay competitive

  • Finistere Ventures and Pitchbook documented 108 angels and seed rounds in 2017, showing a strong upward growth trajectory in agricultural startups

  • A whole range of technological innovations are appearing in the market: companies like BoMill who can sort and identify each kernel in a batch of grain at a rate of 3 metric tons per hour per system, or Vibe using image analytics to identify disease or sprouted grains. There are bio-tech firms (like Pivot Bio), sensor companies (Understory, delivering hyperlocal weather data for farmers), and companies like that are using blockchain to identify and eliminate food wastage, fraud, quality, and safety

  • There are big players too: Indigo secured a $250M Series E investment, pushing their valuation to $3.2 Billion. Originally known for microbial products, they are growing an independent eMarket which may cut out the traditional Grain trading companies. Monsanto are investors in a rival platform, called FarmLead. Existing grain eMarket companies like GrainCorp continue to innovate, adding services to give growers the ability to sell at any stage in the growing cycle, giving them flexibility and protection

  • Cargill and ADM – two of the biggest grain firms, and direct competitors, launched a joint technology venture in October 2018 called Grainbridge ‘to convert data, free of charge to the farmer, into information that will help farmers maximise their profits’.

Data, data, everywhere          

Many of the technology innovations driving change in agriculture produce additional data for firms to analyze. There is streaming information from a range of sensor devices, along with high definition images, weather feeds, point-of-sale and consumer data in fine-grain detail that has never been available before, along with real-time tracking of supply chain elements like trucks, trains, ships, and individual storage units. This tsunami of data floods IT departments, and requires new hardware, software, and expertise to take best advantage.


The Combinatorial Explosion

Trying to make sense (and base decisions) on this data is harder than you might imagine. A combinatorial explosion is building in key functional areas, such as blending and logistics. If you thought the data explosion was bad a few years ago, take a look at how much harder the combinatorial equivalent is to manage:

e.g. a 3-way blend of 100 grain silos with a few criteria which was once several million choices can suddenly have 10 to the power 45 potential options – which is approaching the number of atoms in the earth, for what should be a fairly simple business choice. Of course, operational decisions are still possible, but are a long way from being optimised, and it’s highly probable that current practice is a significant factor off the best choice, and declining fast when compared to decisions that could be made by competitors. This optimisation decline is especially dangerous as it’s invisible, being a comparative change - like thinking your cassette based Sony Walkman is great, as the iPod hits the market. What’s more, logistics is changing too, adding ‘fuel to the fire’ and making it even harder to arrive at optimal decisions in a realistic timeframe.

As data sources and requirements multiply, so will the exponential growth of potential possibilities. What was once a straightforward choice, becomes a decision that is an order of magnitude harder to make, at a time when competition is at its most intense.


Enough of the scaremongering: what’s the solution?

Complex systems are not new. A nest of army ants on the rampage is a complex system, as is a murmuration of feeding swallows, a hive of bees foraging for honey, and even the human body. Nature handles these systems by allowing for the interaction of many entities via simple rules, out of which complex behavior emerges, e.g. individual neurons in a human brain fire and respond to other neurons, even though they have no independent ‘intelligence’ separate from the consciousness of the brain.

Software equivalents are beginning to move into the mainstream, as there are far more agent-capable smart devices (thanks to the Internet of Things), plus AI has advanced rapidly in the last few years, especially in the realm of machine learning.

At SWARM Engineering, we have been combining multi-agent systems with reinforcement learning and applying this to traditional optimizations in grain organizations – such as how to blend the right bins for customer orders, and how to lower logistic costs while taking into account fumigation, seed shapes, critical factors like protein content and falling numbers. Reinforcement learning is especially promising, as it allows a system to simulate a season billions of times, with different decisions, to learn the patterns and choices that deliver the best results. This informs the software agents on how to behave in real-time and makes use of the critical pieces of information from that flood of IoT data. As Google put it, in some of their recent, well-publicized research on reinforcement learning, it allows you to reach ‘…superhuman level without human examples or guidance…’.


Which companies will survive the coming grain revolution?

Organisations that aspire to be new leaders in grain, or to maintain their leadership position, will undoubtedly require more data. They will want new ways to rapidly assimilate changing consumer demands, and the ability to meet growers’ evolving requirements, plus the capacity to influence and be influenced by farmers and other stakeholders. There will be more sensors and better analytics required, new microbial approaches and bioscience, advanced sorting and filtering of grains, along with a corporate sensibility to political and societal factors on sustainability, the environment and climate.

The winners will be flexiblefast, and forward-looking. They will manage a supply chain that can automatically cope with dynamic conditions and frequent disruptions, and that keeps learning and improving from experience. We don’t know any company that has that level of sophistication yet, but we do know of several firms that are on the path. The next grain revolution is already in play.