The Foragecaster project team have a lot to celebrate. After 30 hours of interviews with livestock producers and a couple of deep dives into specific farms, the Beta program of Foragecaster has been officially released on the AgriWebb marketplace.
It’s a milestone for this $6.5m collaborative project between Food Agility, AgriWebb, CiboLabs, FlintPro, Optiweigh, the Queensland University of Technology, University New England, University of Technology Sydney, MLA and NSW DPI.
We sat down with Dr Kenny Sabir of AgriWebb, and Dr Guvenc Dik, Dr Jesse Sharp, and Dr Roy Yang, all of QUT, to get up to speed with the latest in the project.
To begin though, we cast our gaze back to last year, when the Foragecaster team released the Rotational Planner and the Grazing Planner on AgriWebb. These two planners have allowed producers to map out paddock rotations and feed plans in the future.
Since then, the team has been focused on refining the machine learning model livestock growth.
“This has helped us understand trends nationwide,” explains Dr Sabir. “It has also allowed us to focus on regional performance, even considering a particular farm’s historical performance.
“The pasture growth modelling has been looking at identifying different pasture classifications from remote sensing (satellite) data, given how they respond to events such as rain over time.”
The team has also developed a framework for presenting multiple sustainability schemes based on farm data.
They did this by identifying a hierarchical list of sustainability metrics and are now focused on populating them with the relevant data.
“These include Green House Gasses (GHG), Water Access, Biodiversity and Animal Welfare,” says Dr Sabir. “For example, one metric could be the average distance to trough for all your paddocks, where this number can be compared to regional benchmarks.
“We have also been working on detailing the relationship between sustainability, natural capital, productivity and resilience to help predict and mitigate risk when planning,” he adds.
The 30 hours of interviews with livestock producers has given the team the opportunity to understand current practices for planning and their limitations, while designers have worked with the researchers to develop wireframes and mockups of designs to test in front of the producers, which is how the Grazing Planner was developed.
When combined with the 9-month feasibility study that took place in 2023, this project will run for four years. A length of time that Dr Sabir says is “like a lifetime for a startup.”
“For that reason, and to be able to offer value to the producers taking part, we planned iterative releases, starting with the Rotational Planner released mid 2024 with the Grazing Planner released in Q4 2024,” he says.
“This year we will see the MLA Carbon Calculator integrated in AgriWebb and we are aiming for the preview release of livestock growth based on Machine Learning for our Foragecaster Beta customers.”
For Dr Guvenc Dik, a Food Agility veteran of sorts – having worked across two previous research projects – an agile approach gives the team time to refine and validate the product through multiple iterations, helping to guide the research and prioritise focus where it is needed.
“We’ve placed a strong emphasis on developing a minimum viable product,” says Dr Guvenc. “This has been essential to gather timely and critical feedback from industry partners through the quarterly showcases.”
Dr Jesse Sharp believes involving farmers and end-users at every step is vital for a project like this to succeed.
“I have found it really interesting to hear how farmers and other stakeholders place different levels of importance on particular aspects of sustainability,” reflects Dr Sharp. “Their opinions are echoed in the variation you see between the multiple sustainability frameworks and guidelines available.”
For Dr Guvenc, interpreting the interviews that researchers at the University of Technology Sydney have done with livestock managers has helped him better understand and identify the specific needs that the Foragecaster app will address.
“Through their (UTS) work, we’ve gained important insight into the challenges livestock managers face, including issues of productivity, profitability, and sustainability,” explains Dr Guvenc.
“The Beta trials organised by AgriWebb have also given us a direct line to the enthusiastic early adopters that want to explore the latest software and provide valuable feedback. All of this informs our research and development, helping us prioritise areas where focus is most needed.
“It has been very motivating to hear that an efficient decision support tool, integrating cutting edge AI technologies and enabling livestock managers to look ahead in time to make informed decisions today can be a real ‘game-changer’,” he says.
Of the questions the team asked their early adopters, many have centred around their requirements and process for planning and understanding their current limitations.
“An example of this is when we ask producers how far they plan for rotations,” says Dr Sabir. “Some producers say two weeks. When we ask why only two, the answer we would get was, well, they have to do different rotations based on how much rain they get and the levels of the pasture.
“So naturally we’d ask them whether creating a rotation for the various weather scenarios would help mitigate their risk and plan better for workforce management, and the answer, after a quick think, is yes that would be helpful.”
This feedback, Dr Sabir says, has then allowed the team to build features into the Foragecaster app they’d previously not considered.
“We found that producers were sceptical about weather forecasts,” he says. “However, they trust historical weather patterns.
“So, we worked on providing an option that would allow farmers using the app to compare current conditions to historical years. This allows them to relate to the planned rain through their ‘lived experience’ - for example, the current conditions are similar to 2011 so we could use that as a template for forecasting,” explains Dr Sabir.
If it wasn’t already obvious, this project has collected an enormous amount of data, which really pleases Dr Sharp. “As a data scientist, I am excited by the wealth of information that AgriWebb has collected to drive research in the Foragecaster project,” he says.
“Farm level data means tailored insights for farmers. From a sustainability perspective, I am excited to develop quantitative metrics that farmers can use to measure and track their sustainability over time, see how they stack up against regional benchmarks, and identify opportunities for improvement.”
The AI part of the project is what Dr Roy Yang is most excited by. “Foragecaster represents a unique opportunity of studying the development of an AI-enabled decision support system which encompasses various algorithms and complex data crunching,” says Dr Yang.
“I am excited to be able to contribute to ensuring the reliability of such a system from a data quality and architect angle.
“Moreover, the user-facing nature of Foragecaster also excites me as an early career researcher, as it gives the opportunity to show how research may be translated into real-world applications and bring benefits to farmers,” he adds.
For Dr Sabir, this project represents an opportunity to understand the relationship between farm productivity and sustainability metrics to help create more resilient farms.
“To give you an example, with the anonymised sale and purchase records, we can identify farmer cohorts from the 2017-2019 drought that bounced back quickly, slowly, or the ones whose farms did not cope resulting in farm-sale identified in the subscription churn reasons.
“We can then correlate these cohorts to their sustainability metrics to understand if there are any trends. For instance, what practices did the highly productive farms have in common, such as pasture biodiversity, percentage of natural grasses, or destocking early?”
Dr Sabir says his overall goal for this project is to provide farmers with the confidence to trust the system. “If a farmer using the Foragecaster planning tools wants to know the reasons behind a recommendation or prediction, we can provide them with the data to justify it.
“That confidence will ultimately help mitigate risk and improve efficiency,” he concludes.
AgriWebb Producers can join the Foragecaster Beta, available on the AgriWebb Marketplace.
New producers can sign up for an AgriWebb 2-week trial where the support team is ready to help get you set up to be effective.
