30 MHz is an Amsterdam-based software company

who set out to build an analytics and management platform for smart devices.

They never dreamed they would end up in agriculture.

We started with a cloud infrastructure model platform.

We developed this on AWS.

I'm Jasper Geurtsen, Co-Founder and Head of Hardware Department in 30 MHz.

The first few years we focused on building the product and finding our industry.

My name is Flavia Paganelli, I am Co-Founder at 30 MHz and former CTO.

We knew that we wanted to combine our strong AWS infrastructure

with wireless sensors to digitalize some process,

but we didn't know what process yet.

30 MHz found the ideal match for that combination

in the agricultural industry, specifically greenhouses and indoor farms.

We realized that we were culturally a very good fit.

And then, we decided to focus on building the product knowing that we would add value to the space.

With the global population predicted to reach 9.1 billion by 2050,

the world's food production needs to increase by at least 70%.

Meaning more farmers grow more food for more people

while not taxing our planet's precious resources.

We set out to make a data-driven technology for any farmers in the world, for any states.

We knew when we came into the agriculture field that we had technology

that could really help low-tech farmers bridge the gap with high-tech farmers.

The original 30 MHz platform culled and analyzed data

from infield monitoring devices, allowing farmers to remotely monitor

and manage much of what happened in their greenhouses.

What you see is that farmers are cultivating ever-increasing plots of land

and with the script of predictive analytics, we can help them

manage these areas more effectively.

I'm Daam Rutten, I'm a data scientist working for 30MHz.

In the early days, teams had to troubleshoot outages, loss of data,

and found the monitoring system difficult to scale.

So we had to re-design part of our architecture to make our gateways

run autonomously.

But we saw this as an opportunity to drop some of our software

and start using AWS IoT.

With the support of the AWS prototype team,

we managed to integrate AWS Greengrass into our gateways,

and this way we loosened the boundary between the edge and the cloud network

and really make this flexible, and we get additional features

like offline buffering, so when the connection drops,

then Greengrass will store the data for us, and when the connection is recovered,

it will send out the missing time.

My name is Amir Majlesi.

I look after AWS prototyping activities in development at Nordics.

We collaborated with the 30 MHz team in a hands-on rapid prototyping engagement,

helping them modernize their architecture and build

a serverless smart agriculture solution.

We co-developed a working prototype which was customized to their business environment in just 3 weeks.

The new prototype solved the device management and connectivity issues,

and helped 30 MHz automate the provisioning of gateways.

It also added capabilities such as over-the-air updates

and advanced security measures to protect data.

In 2019, the team participated in the Autonomous Greenhouse Challenge,

sponsored by Wageningen University & Research.

The goal of the Greenhouse Challenge was to autonomously manage

a greenhouse for 6 months. We were cultivating tomatoes.

The experience pushed 30 MHz to look closer at how they could bring this type of data analytics

and machine learning to real growers.

And I think it was the introduction to the challenge that also introduced us to

the possibilities for autonomous growing for actual greenhouses.

30 MHz sought to develop a solution that combined the data monitoring software

and the data collecting hardware with machine learning.

This machine learning technology would enable growers to make real-time decisions

on climate and irrigation, creating optimal growing conditions for crops.

The 30 MHz team had highly specialized expertise in digital agriculture and data science.

The port of engagements really builds upon these existing skills

and enabled the team to develop a new machine learning pipeline with Amazon SageMaker.

The AWS team helped 30 MHz analyze the data, build and deploy the machine learning models

for inference, and automate the machine learning process from beginning to end.

We implemented the architecture in 2 parts.

Model training and inference.

First, we fetched the existing 30 MHz codes from their code repository.

We trained this model with Amazon SageMaker, using specific data sets from each grower.

Once training is done, SageMaker stores the model artifacts in an S3 bucket.

The next step is to evaluate the model performance to check if the accuracy is sufficient enough.

After it passes, the model is deployed.

Once deployed, the model is now ready for inference with incoming sensor data

collected from customer fields using AWS Greengrass.

So one of our customers is an IPM consultant

and he has been using the 30 MHz platform to really look into the data

with grower groups on situations

where pest and diseases might increase and analyzing this micro-climate data,

including the crop log data to really see situations

or conditions that may explain these kinds of situations.

For these grower groups, understanding and visualizing this data

on the disease management is really helping them to push back the infection.

They not only save money, but it's maybe more important that they're really able to

leave out some chemical solutions that might otherwise

be necessary to really push back the disease.

So we envisioned our technology being available everywhere in the world at a fraction of the cost

of climate computer systems.

And this will dramatically lower the costs of running a greenhouse.

This can happen everywhere in the world.

The 30 MHz prototype is currently being tested by 8 pilot growers,

with the hopes of scaling up to hundreds, even thousands, of growers in the near future.

The insights and information that this technology gives to our customers

can be life-changing.

The collaboration with AWS prototyping has been key for this,

because we have been able to build and test fast

and also create solutions which are scalable and reliable from the beginning.

The 30 MHz team continues to look to the future, refining and expanding the capabilities

while keeping the technology easy to use.

I think one of the things that we've experienced is that it's possible to have

real-time autonomous control.

The main challenge, however, is to incorporate real-time feedback from the crop,

which we still aren't able to measure, or at least measure to the extent that we'd like.

Today, 30 MHz has 35 employees, and services over 330 customers

across 30 countries and counting.

30 MHz wants to be the leader who brings the infrastructure

and information to growers to allow them, and the whole industry, to advance

at a much faster pace.

And in this way, they can make better use of resources and feed the world.