AI and Machine Learning for Accurate Predictive Analysis

This week, CannabisTech took the opportunity to speak with Ally Monk, CEO of MotorLeaf, about this crystal ball technology designed for tomatoes, cannabis, and anything else produced under the sun… (or grow lights).



Imagine having the ability to hire an entire team of expert agronomists to work around the clock - monitoring and controlling every aspect of your grow - predicting yield, anticipating disease, and making suggestions to improve the outcome of each crop. Hiring this kind of team would be costly and impractical for most cannabis facilities. Fortunately, MotorLeaf, a company based out of Montreal, Canada, discovered a way to use advanced AI and machine learning to build Agronomist AI – a revolutionary virtual agronomist who can do what no human can – process millions of points of data to predict the future.

iRobot Meets Farming

“The best way to think about what we are doing is to give every grower the best possible chance at success – using data as the thing which drives decisions, not just experience and gut instinct,” Monk explained. While human instinct and experience are invaluable tools, as with any industry – data is king. Many systems currently exist for collecting the data and displaying stats in dashboard-style layouts, but MotorLeaf analyzes the data collected to make accurate predictions for harvest yields, as well as, disease development.

Agronomist AI is a virtual agronomist who can perform millions of calculations while you sleep.

A Tool Not a Replacement

Monk pointed out Agronomist AI is not intended to replace your staff, “you need the human touch, someone to actually look at the plants, touch them, and understand what is happening in the greenhouse.” Preferably, the software should be used as a network of experts to help your master grower and staff perform calculations and tasks impossible for the average human.

Integrated AI and Machine Learning

Regardless of where the data originates, MotorLeaf Agronomist AI can use it. “Everything starts with data – you can’t run an algorithm or do machine learning without data. Whether the data comes from our hardware, another manufacturer or existing system, or whatever you have in place – that data makes its way into the software that we build,” Monk said.

The Grow Journal

MotorLeaf brings your greenhouse or indoor crop data to life within the Grow Journal. Providing growers and owners the ability to compare data sets to find the correlations between yield and any number of criteria such as humidity levels, light spectrum, or air temperature. Plus, through the integration of human data, like photographs and notes, tracking events or conditions which could impact yields allows cultivators to compare crops more sensibly. “You want to look back at what you were doing and why you’re doing it to understand why that yield was so much better or worse than before,” Monk suggested.

More than a Dashboard

Maybe the most impressive aspect of MotorLeaf is its ability to learn from the data and predict yields and the potential for disease. With trials showing more than 90% accuracy to predict infestations like White Fly, and many others, Monk believes the predictive capabilities of the software are invaluable for indoor cultivation facilities.

“Our mentality is that ‘AI’ is not a buzzword – it is what we do,” Monk proclaimed, “The real value is in the machine learning.” Explaining the difference, Monk used a robot vacuum in a business lobby as an example of artificial intelligence. The robot vacuum does in less time; something humans can do on their own, but choose not to, so they can focus on more rewarding tasks. Now, if that same robot vacuum also collects data points, such as distances between certain points, busiest times of day, and weather conditions to calculate the best time and most efficient way to vacuum the space – that is machine learning. “It actually uses the data it collects to learn a better way to do things.”

Agronomist AI looks at measurements like pH levels, irrigation schedules, lighting spectrum, and other data points, comparing them to yield, and then learns from the data it collects. Looking for the best possible way the data shows, the software analyzes millions and millions of variations before providing the answer.

“This is exciting for any kind of crop, if you understand how to calculate yield – it’s not a ‘number’ – It’s the understanding of why the number will be what it is going to be and if you know that, then you know what “levers to pull” to make necessary changes. Humans can’t do that. You can’t make that kind of analysis of that kind of data, calculating every minute of every day.”

Agronomist AI can.

The Results Speak for Themselves

Benchmarking their software against the human, Agronomist AI is reducing errors in yield prediction by an average of at least 50% with many clients reporting 75% or more. “If you’re more than 15% off (in your predictions) that is a significant error. We reduce the error and the severity of the error – essentially smoothing out that line,” Monk explained.

On the Horizon

MotorLeaf is actively involved in open agriculture API development; understanding client data needs to be able to travel between service providers so growers can utilize their data. Additionally, Monk mentioned they want to work with lighting providers as well. “Light is a key data point for learning about other areas of the operation, and the data needs to be incorporated.”

While the company may have started in 2015 with two guys in a pub scratching ideas on a piece of paper, Monk anticipates company expansion in 2019 to double the number of employees the company currently maintains to provide services to a wider variety of clientele. Talking about their technology roadmap, Monk stated,