Contextual comms in agriculture

An Interview with Bruce Rasa, AgVoice

The frontier of IT isn’t necessarily found in Silicon Valley, at least not now its fruit orchards are long gone. Since necessity is the mother of invention, those who continue to harvest from capricious Mother Nature have found an ongoing need for innovation.

I interviewed Bruce Rasa, CEO and Founder of AgVoice, based in Buford, Georgia. They are applying wearable technology, hypervoice, AI and Big Data in the farms and fields of America and beyond.

MG: Who are you, and what problems does your company, AgVoice, help to solve?

BR: My background is growing up on our family farm in western Missouri. My uncle’s farm has the second largest apple orchard in the state of Missouri, which sells about half its annual produce to Walmart, who in turn buys 24% of all food in America.

I cut my professional teeth at IBM, NCR, and then AGCO Corp, who make tractors and farm equipment. In that last role I was in charge of global precision agriculture marketing, using telematics and other data capture from agricultural machinery.

AgVoice is a start-up focused on agriculture voice capture, with a view to improve process efficiency and in-field productivity of agriculture professionals.

To understand the problems my company is trying to solve, it helps to look back in history at that apple orchard supplying Walmart. My great grandfather, August Rasa Sr., planted the first trees in 1922. Since then my grandfather and his three sons have been tending the orchard, which now includes 40,000 trees over 175 acres.

Three years ago, Walmart created an initiative to improve sustainability and compliance. Now, Walmart’s need is a legitimate societal demand – a very straightforward one. Consumers ask Walmart every day: Where did my food come from? Was it grown in a safe and responsible way? How were the workers treated? These are all reasonable questions in the pursuit of food safety and traceability. This creates a new reporting challenge.

Big Data is also coming to agriculture, first in the US, UK, and Australia, but soon it will reach many developing countries. There is a growing range of sensors, almost all derivatives of mobile phone industry technology. Indeed, the “Internet of Things” is old news for us, with combine harvester machines and sensors on animals. The agriculture industry has had satellite-guided steering for the last 15 years.

So the issue is the overwhelming amount of data coming in, which is incompatible and fragmented. What we lack are the tools for data integration to meet those paperwork and compliance demands.

How are your customers responding to this need?

There has been a need to keep records for hundreds of years, tracking agricultural production. I have a photograph of my grandfather’s pocket notebook from 1937. It’s a classic farmer’s pocket ledger, easily controllable and very personal. Farmers knew their operation well, as they lived in the middle of the business.

Nowadays we have touch screens in tractors and combines for data entry. They automatically drive more accurately than any human can, and are festooned with sensors for guidance and crop yield measurement. The job of the person in the cab isn’t to ‘drive’ as such, but rather to manage the performance of the (sensor and data-driven) harvesting process.

Increasingly those involved in farming also have a smartphone, with penetration estimated by one farm journal magazine to have reached 86% in the last two years. Around 60% of the US farming professional worker market has an Apple device (iPhone or iPad). That choice is not just a love of the brand, but because it is a complete solution that ‘just works’ for our kind of industrial use cases – where there is often no IT support around.

So the needs of our customers are centred on exploiting their investments in personal IT in the form of smartphones and tablets, in order to better perform data capture and integration.

How are you helping them addressing their evolving needs?

We have been experimenting with leading-edge wearable technology to enrich this information gathering process. For instance, I was accepted in the Google Glass Explorers program. We put the technology in the hands of 800 people over a 14 month period from 22 different countries, as well as attending many international agriculture conferences.

Users were clear: if we can work hands-free, then we need it. That means they love voice-driven interfaces. Yet with Google Glass in its first beta version, people had to tap their temples too much. That doesn’t work if your hands are dirty or greasy, or if you need to have both hands on (or in) an animal. So to succeed you need a more purpose-optimised device than the general purpose offer from someone like Apple or Google.

The secret sauce we develop is how to apply these learnings whilst exploiting the existing device investments. We are an experience integrator, looking for the best raw capabilities to incorporate for outdoors industrial use, especially where voice can be used as the most practical interface. Our business generates IPR around these new use cases.

What is an example of a new use cases of personal IT in agriculture?

Our first target market is not farmers, but those who support them: ‘plant doctors’ who inspect the health of the crop. This kind of agronomist is called a ‘crop scout’. Their job is to identify anything that affects the yield, in terms of volume or quality.

An agronomist like this often carries 5-7 objects into the field: a shovel, small bags, scales, a soil sample collection kit, and measurement tools. They might have a camera, with proper lighting and a background board for imaging plants.

They are looking for new problems, as well as outliers from the crop norm. For fruits and vegetables this can be a labour intensive task. If you don’t catch a disease issue quickly, it can have an extremely high cost in a couple of days. So to be truly vigilant you need lots of tools, often more than can easily be handled by one person.

You also want to record your thoughts and insights. Yet very few tools work hands-free and are voice driven. So when they get back to their vehicle, they have to write up notes. This might mean tapping on a tiny smartphone screen in the vehicle, or typing away in a motel at night. Having been in seven grain fields in the day, the worker can’t remember with high accuracy what was in each field.

In the early 1990s GPS was a total game-changer for agriculture, as everything effectively became geo-referenced. It’s a wonderful platform, but we need more. We continue to interview people like researchers and geneticists, who are tracking plant varieties over multiple years. They may trace the efficacy of a product on the crop to understand if and how it works. But they don’t have an extra hand to hold an iPad (in an industrial-grade case), nor a way to simultaneously record information with high accuracy whilst doing their work.

You say you have my hypervoice presentation plastered on your office walls. Why is hypervoice so relevant to agricultural tech?

Hypervoice is relevant to many industries, but one thing is nearly unique to agricultural technology: we require an offline mode. It’s no use having everything in the cloud, since there are many places where there is weak, expensive, or no cellular coverage. Hypervoice fits with that offline use need, as it ties voice to its context.

I also very much like the hypervoice term and construct. It is important because it helped me to crystallise, in a straightforward way, how much is possible with voice. The hypervoice concept shows why voice isn’t close to dead. The variety of different capabilities means it has a great deal of latent potential.

As a result we are looking to see what we can take from the smartphone realm and general digital infrastructure into to our vertical industry. Our experience with hypervoice suggests that there is a fascinating roadmap ahead for both consumers and general business use.

The catch is that whilst voice data has high economic value, it is also very personal. It needs a higher degree of data security than before due to this combination. It is insufficient to just build an app, throw it on the app store and wish your users good luck.

What else have you learnt from applying hypervoice technology?

What we are doing is looking for those select pieces of the hypervoice concept that could meet near-term needs: it’s like sci-fi, but we want customers to adopt right away. That means we have to focus carefully on specific pain points.

That gets them started with the paradigm and then we can grow their relationship with the technology. It’s not sufficient to announce “hypervoice has 10 major principles” and then try to pack them all into one ideal system. Instead, just pick one or two and put those into action.

We are finding those select pieces, and in parallel discovering a fascinating wellspring of people with innovative voice technologies. There are lots of platforms and APIs, but they are designed for an office or desk worker, without thought to industrial use cases. Yet this is the era of contextual computing (or ‘hypersense’), where voice is the gateway for everyone and everything.

We humans are multi-modal, but the technology is still catching up. There are many different ways to extend our capabilities with media, gestures and location. But voice is that Holy Grail – if you can make it practical enough.

Sadly, we’ve overstretch with Siri, as it lacks consistency and expectations have been wrongly set as a result. Instead we need to focus on a narrower dictionary and fewer use cases, each with better accuracy.

I believe from what I have experienced so far in industrial use cases that intelligent voice has to be purpose-optimised. This is in contrast to the trend for general IT tools like phones, tablets, and watches, which are each designed for mass consumer markets. They just aren’t right for industrial uses.

For example, my brother is a farmer, and he is 6’5” (195cm) tall. He bought a new high end iPhone 6+ smartphone, and put it in an industrial-grade case. Now it’s too big even for his pocket! These generic hardware platforms are not as suitable or safe as they need to be.

Where do you see this sector going in the next few years?

I believe you are going to find hypervoice fascinating in both developed and developing markets. The world going from 2bn to 4bn smartphones in 2020, per the VC firm Andreessen-Horowitz. The low-end entry level phone is where the aggressive growth is now.

My hunch is that smartphones and wearables, plus hypervoice will have an outsize positive impact on rural economies. These farmers are going to have better access to markets, pricing, growing info for yield and safety. I absolutely see a huge positive upside, worldwide.

Furthermore, I don’t see any let-up in consumers wanting to know more about their food, for reasons of environmental sustainability as well as personal health. The agriculture industry has a responsibility to honour its ‘societal license’ to take care of all its stakeholders. That means we need a quantity and quality of food grown in a visibly sustainable way for the farm economy.

For that to happen, mobile tech will have to become ubiquitous in agriculture. Farmers are already on the phone all day long. That’s just the beginning. I know of a specialty company that writes software for crop scouting, based in Iowa in the heartland of the US. In their area, they don’t know of a single farmer who does not have an iPad. They sit there in the local coffee shop, looking at fields and crops like a soccer match.

So there’s an amazing upside, it just means a lot of challenges to adapt the infrastructure to meet our vertical needs.

In meeting those challenges, what do you know now that you wish you had known at the outset?

The pleasant surprise is how rapidly voice-to-data technologies are evolving. They are not perfect, but that trajectory of improvement has increased my confidence that they have wide utility in a business environment.

Much can be done with voice recognition now, within a specific use case. One reason it is hard to broaden its use is simply background noise and having adequate weather (i.e. not too windy or rainy). You need an excellent physical place to capture good quality voice.

From a hands-free use standpoint it is clear that the perfect solution today is a phone in your pocket or on the side of your belt, with one or more separate microphones for audio capture. These could be clipped onto a vehicle or the person. This microphone positioning is proving to be a tough problem to solve, since we are so accustomed to holding a rectangular brick smartphone.

The evolution to a fully wearable phone is taking time, and I wish the battery life of wearables evolved as fast as our customers’ imaginations. The core smartphone can satisfy many needs, but there are many very specific ones where it falls short. For instance, someone climbing on a grain storage silo must work with two hands free for safety.

At the moment those workers simply can’t hold their phone and do their job. However, as battery technology gets better we feel that many latent needs will surface. This is similar to how in the 1990s around 20% of all workers were found to have push-to-talk needs.

We are seeing a high potential for a subscription service (and not just one-off app sales) to keep data for 10 years for certification and audit purposes.

For us, the challenge of outdoor use cases remains the offline mode, since this is not a main focus in telecoms technology and voice platforms. That makes our vertical industry a bit of an ‘odd duck’. That said, the kinds of access technologies being pioneered by Google and Facebook to reach emerging markets will offer radically more data-efficient services, plus offline modes. So I am hopeful of the offline use issue being resolved.

As this happens I foresee an era of digital agriculture with a stunning upside potential, as information technology and food systems combine. We have context, we know the customer.  They have a legal right to the data and want to understand the macro trends. I have never met a farmer who wants more paperwork. Hypervoice technology becomes more believable and useful by the day in solving that issue.

To get in touch with Bruce, you can contact him at You can also sign up for the AgVoice newsletter at

For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.