156. The Fusion Frontier

The Fusion Frontier

Do you remember the Adobe Encyclopedia Atlantica ad from a few years ago, where a baby keeps hitting the “Buy” button again and again?

It triggers a chain reaction in the supply chain with manufacturing, factories, shipping, and ultimately ends up launching a thousand ships.

It was the modern day version of Helen of Troy, whose face launched a thousand ships in war!

Just like the fun and fictional Encyclopedia Atlantica ad, software will continue to play a larger and larger role in the future within our supply chains. The different parts of the ecosystem will be more and more connected through a layer of software and data.

There are many such examples today.

Doordash - You sit on your couch at home, go through a few screens, and within a few minutes food shows up at your place.

Your phone and your car have more computing power and software on them, than was used for landing the first man on the moon.

With services like Uber, your smartphone is the “main thing” now and the car is an accessory.

Image from a series of essays written by Venkatesh Rao (“Breaking Smart”)

Higher levels of connectivity creates networks, which are also primarily defined in software today.

Again going back to the “Breaking Smart” series,

We don't just live on a networked planet. We live on a planet networked by software, a distinction that makes all the difference. The software-networked planet is an entity that can exist in a continuous and coherent way despite continuous hardware churn, just as we humans experience a persistent identity, even though almost every atom in our bodies gets swapped out every few years.

Even at a micro level, almost every piece of hardware has an element of software now. Smart refrigerators, cars with digital keys, software enabled weighing scales, WiFi and/or Bluetooth toothbrushes, and many such examples.

The software enablement is changing how a piece of hardware is designed, tested, sold, operated, updated, and services.

Traditionally when you get a piece of hardware, you can replace certain parts or services to continue to get value from the equipment, and get more life out of it. You still have to do it today, but now software updates can change the behavior and capabilities of an existing piece of hardware, without making a hardware update.

The Fusion Frontier

Agriculture and food ecosystems are by definition industrial and asset heavy in nature. Think of all the equipment, land, transportation equipment, factories used for R&D, as well as production. Agriculture is more complicated due to the presence of natural resources like seeds, soils, weather etc.

How do the principles of combining software with hardware apply in the case of agrifood systems?

A new book (Fusion Strategy: How Real-Time Data and AI Will Power the Industrial Future) from Professor Venkat (Boston University) and Professor Govindaran (Dartmouth) addresses this question for industrial sectors, and asset heavy industries. They define it as a fusion strategy and define a new frontier called the “Fusion Frontier”

The fusion frontier is a future state in which industrial products are infused with sensors, software, and real-time telematic functionalities through the seamless convergence of physical and digital domains. This allows for enhancing the productivity of industrial assets and delivering personalized approaches to solving business problems with algorithms that make use of data observed across different settings. In this frontier, industrial companies excel not by simply designing and delivering brilliant machines but by ensuring that their machines satisfy the specific needs of individual customers.

If we want to look at a very recent real-life example, Tesla has recently moved their FSD (Full Self Driving) mode from beta into offering a 1 month free trial to everyone. This was done through a software upgrade and turning on the FSD bit on compatible models of Tesla.

Tesla is able to do so because they have already been testing FSD functionality, and collecting massive amounts of data from actual users and actual road conditions to continue to tweak their FSD artificial intelligence models.

Back in 2019, Elon Musk made an astonishing claim for Tesla vehicles.

Tesla cars, he said, would go up in value, not down, after purchase. The reason for that is Tesla’s full self-driving capability that, Musk has said, requires only some additional software updates and regulatory approval before Tesla vehicles on the road today will become fully independent.

He repeated this claim as recently as June 2023.

“You can think of every car we sell or produce that has full autonomy capability as something that in the future may be worth five times what it is today,” he said in the company’s third quarter, 2023, earnings call.

Even though this did not turn out to be true in the case of Tesla cars, and Tesla actually dropped the prices of their cars in response to competition, it is an interesting idea. Software updates can add new capabilities to your existing piece of hardware, and increase the utility or useful life of that piece of hardware.

Faster product development through fusion products

The always opinionated (!) Walt Duflock recently posted on LinkedIn, (Highlight by me)

As pointed out by Walt, Tesla has used their rich dataset which they have acquired through actual usage by customers to further enhance and improve their FSD capabilities. This is possible as Tesla has pushed against the fusion frontier to architect their products to be a combination of digital and physical capabilities enhanced through the use of connectivity and artificial intelligence.

Fusion Strategies

In the book, Fusion Strategy, the authors lay out four different fusion strategies and the steps needed to execute on those strategies. Each of these strategies require a richness of high quality diverse data sets, incorporation of fusion principles in product design, leveraging the latest tools like artificial intelligence, collaborating with other players within the ecosystem, and evolving their business models.

Image from Rhishi Pethe’s copy of “Fusion Strategy”

What does this mean for Agrifood?

The agriculture sector obviously uses equipment to perform different operations on and off the field. Equipment has a natural structural advantage to push the fusion frontier, as it is easier to combine digital and physical in the case of equipment, compared to a product like seed, or chemicals, or some biological products.

Let us look at a few examples, with different flavors of fusion strategies

Deere (Fusion product)

Due to this, we have seen equipment companies like Deere, and specialty crop robotics companies like FarmWise, Burro, Monarch, and others to really push the fusion frontier with integration of sensors, connectivity, software, and artificial intelligence with their hardware products. Deere is a great example of how they have redesigned their whole end to end process of product ideation, design, delivery, service, and updates done to their equipment.

With a capability like see & spray, or harvest assist on combines, sensors and telematics are heavily integrated in the entire product lifecycle. The product design has to include ways in which data collected during the actual operation of the equipment in the field can generate new insights to continue to improve the product offering for current and future generations of the product.

For example, as more and more Deere machines are equipped with cameras and edge capabilities, they will learn how to improve the quality of their AI models by understanding different crop types, weed types and weed growth stages, soil background, lighting conditions etc.

This data (with the right legal and privacy considerations) can be used to not only improve existing models but also to build new models much more efficiently. These software enabled hardware devices become learning systems of their own, and provide a competitive advantage to a company like Deere compared to others who are not pushing the “fusion frontier.”

If an aftermarket approach is followed, this is an example of fusion strategy by enabling existing physical systems with additional digital capabilities to help accelerate their fusion roadmap.

FarmWise (Fusion product bleeding into Fusion services)

I recently spoke with Tjarko Liefer (CEO of FarmWise) and Greg Chiocco (VP of Product of FarmWise), about how data collected during product development and actual operations helps them improve their existing products.

For reference, FarmWise is a robotics company currently within the specialty crop space with a mission to “Enable farmers and ag equipment OEMs to reimagine every farming task with machinery that can perceive, understand, and act at the plant level in real time.”

According to their website,

FarmWise® deploys state-of-the-art AI and computer vision weeding technology to help farmers achieve their productivity and profitability goals.

Their product Vulcan is a classic definition of pushing the “fusion frontier” as it combines hardware, sensors, cameras, and artificial intelligence in a seamless manner.

Due to the data collected by FarmWise during the training process, they continue to get new data to improve their models.

For example, if they are training a weed model for a new crop type, the data about new soil backgrounds, lighting conditions, furrow architecture, other plants etc. actually helps them improve their underlying base model (also called a foundation model). This improves the existing model for an existing crop type, without collecting additional data in that crop type.

As Greg Chiocco posted on LinkedIn, the continuous collection of data has helped FarmWise improve their foundation model, which improves their existing crop specific models

FarmWise is an example of a fusion strategy from Day 1. This is especially true for newer robotics companies, who can take the benefit of new approaches from day 1.

FarmWise’s software models are tightly coupled with their hardware, and so provide a weeding system, which is a combination of its hardware, sensors and vision systems, and artificial intelligence models.

These companies have architected their product as a fusion product, set up processes to get usage data, and use it in combination with tools like artificial intelligence to continuously improve their existing products, and increase the speed at which they can bring new products to market in the future.

Swan Systems (Fusion system)

Swan Systems is a provider of irrigation software for different sectors including agriculture. Understanding your irrigation needs requires a variety of sensors and probes, remote sensing data like weather, crop models for water usage etc. Swan Systems is designed to work with a large number of hardware devices. The combination of these hardware devices (combined with other data sources) and the software from Swan Systems, creates a fusion system, whose functionality is enhanced due to the presence of software.

According to their website,

Unlike other irrigation management products, SWAN Systems is designed to work with any hardware. This makes it very adaptable and versatile, enabling you to manage irrigation at the whole farm level in one platform; even with a mixture of infrastructure types.

The example of Swan Systems is similar to Uber. The Uber app almost abstracts out the type of vehicle used to provide you transportation services. In the case of Swan Systems, as a user you are almost agnostic to the hardware being used, as long as the hardware performs as designed / planned, and in many cases hardware is not even required to manage your irrigation needs

Swan Systems is an example of pushing the fusion frontier through collaboration with ecosystem players.

Structural Advantage

As I said before, equipment and sensor companies have a structural advantage, when it comes to fusion strategies to create new value for their customers. It is harder for seed and chemical companies, as they can incorporate fusion strategies in product development, but it is much harder to get data during use, and they have to rely on other systems and data modalities to get usage level data. (For example, satellite data to see performance of a particular crop, Yield Monitors to see the performance of a hybrid under different conditions etc.)

Due to this, I would expect to see more collaboration between equipment companies and other agrifood organizations, as equipment and sensor companies have pushed the fusion frontier. There are already a few examples like Bayer’s partnership with Microsoft, its use of the FieldView drive, and its partnerships with other OEMs. Syngenta Ventures has invested in GreenEye, and Syngenta has close relationships with other OEMs.

So the big question for your organization is whether you are pushing the fusion frontier or not.

If you are not, you run the risk of being left behind compared to other organizations who do push the frontier.