Software solutions like Cropster and fabscale are leading the digital growth of the coffee industry – with features ranging from data collection and analysis to predicting roast outcomes.
Artificial intelligence (AI) was once considered a piece of science fiction, and while we’re a long way from machines taking over the world or passing for people, it has taken leaps and bounds in recent years. Machine learning, put simply, involves a computer using a large existing database to make decisions or predictions about something happening now or in the future. Through Bean Curve Prediction, Cropster has put this technology to work for the coffee industry.
“Bean Curve Prediction provides immediate benefits to all of our customers when they roast. It makes sure they stay on track during production roasting, leading to better consistency and less wasted batches, saving time and money,” Cropster Roast Product Manager Lisa Gringl says.
Bean Curve Prediction begins 60 seconds into a roast and appears as a dashed curve extending past the running bean temperature curve. Gringl says it is accurate within 1°C for two minutes into the future and 0.5°C within 30 seconds. She adds that Bean Curve Prediction shows its true potential before production, during sample roasting where profiles and recipes are still being set for new green coffees.
“With this tool, you need less sample batches to develop your new product. When you get a new coffee, you’re usually starting with a blank slate. Bean Curve Prediction has learnt from thousands of anonymised roasts, providing a tool to immediately see where that roast will go. It even learns from new roasts and what you do at the moment, so it gets smarter,” she says.
“A third benefit we hadn’t considered during development, and discovered during beta testing, was to train staff. If a new roaster joins the team and you want to teach them how the coffee reacts to specific changes, you can do that with the curve prediction because it updates every second and takes every gas or airflow change into account.”
The development of Bean Curve Prediction began several years ago, when Cropster approached several of its customers to gather data and insight from the roasting process. Cropster was overwhelmed by the interest in the project, with many customers wanting to find correlations in their own data.
A team of data scientists and mathematicians uses that data pool as an “incubator or test bed” for new developments in AI for Cropster. While Bean Curve Prediction was Cropster’s first AI program in the market, Gringl says it was not the first in development, with many more on the way.
“We try to pose questions or fulfill tasks that are beneficial for Cropster users. As an example, we’re working on predicting inventory, so ‘when am I going to run out of a green component?’ This is something we can predict,” she says.
But AI is only one way Cropster works automation into its solutions for the coffee industry. With a product and supply chain as complex as coffee, Gringly says automation can be implemented in many ways and stages.
“AI is about finding processes that are standardised and repetitive, then breaking them down into small pieces or concrete tasks, and training a system to repeat those tasks,” she says.
“At Cropster, we focus on three areas of automation: one is machine automation for roasting, another is workflow automation in the field and for traders in the coffee lab, and the third is workflow automation within roasteries.”
For workflow automation in the field, Cropster launched QR codes several years ago that can be shared alongside coffees, and due to their positive reception, has continued to widen these avenues for sharing data. Using the Cropster Origin app, producers can capture field data and automatically share it worldwide. On the other end, traders and roasters can cup coffee and provide immediate feedback to the farmers. Direct access to this information saves both ends the time of collating and sending this data to the other.
“Getting coffee from origin to the final cup involves many partners and passes through a lot of hands, so it’s very important that information is shared with a lot of efficiency and accuracy,” Gringl says. “In the beginning, just a few early adopters wanted to use the feature. Now, as they get more tech savvy, we receive daily enquiries from customers that want this workflow automation.”
Inside the roastery, Cropster has developed automated processes to improve production planning as well as workflow. The Order to Roast program, for example, collates coffee orders and uses that information to help create a production schedule. This information can either be entered manually or through integration with eCommerce programs like Shopify or WooCommerce. Gringl says this type of integration is important to workflow automation, and a key focus of Cropster going forward.
“With large customers who use [enterprise resource planning] platforms, if you can connect systems, that means less or no manual data entry, saving time and reducing the chances of errors,” she says. “We work directly with these customers and their IT departments so they can integrate in Cropster via our [application programming interface] (API).
“The goal of all this technology is for workers in the plant to get more information to perform their key tasks and use their time more efficiently. Reducing those repetitive tasks will also reduce the errors that take place in a business.”
For more information, visit www.cropster.com
HEAD IN THE CLOUDS
Going hand-in-hand with automation is the other Industry 4.0 concept of digitalisation. As coffee processing equipment becomes more advanced, the level of data they produce increases. Seeing a gap in the market, Cropster partnered with Probat in 2019 to launch fabscale, an independent cloud-based solution that collates this data from throughout the entire roasting production process and offers rich analytics and notification solutions to increase efficiency.
CEO Christian von Craushaar tells Global Coffee Report there is growing demand from large scale coffee roasters for smart technology that improves workflow and decision making.
“We’re seeing a whole new set of technology become accessible, and it’s leading to a digital transformation process in industrial production,” von Craushaar says.
“The food and coffee industries have been slower than many other forms of manufacturing in adapting these technologies. We see a need, and fabscale wants to be the first comprehensive, entire-plant solution specialised to the coffee industry and production on an industrial level.”
Fabscale’s first product, a dashboard-style display of data, includes a reporting tool, where detailed reports can be produced at the push of a button once important factors have been selected. Von Craushaar says this dashboard provides a base, on which fabscale will build and develop other modules to further automate and streamline the production process.
“One of our upcoming solutions will monitor the condition of the roaster, automating the maintenance component. Preconfigured condition monitoring tells the client, based on the lifespan, overall roast time, age, and usage of a component, what and when maintenance tasks need to be done,” he says. “We’re also creating an API so customers can access their own data and implement it in any solutions they’ll need or use certain types or sets of data somewhere else.”
Von Craushaar is looking forward to seeing how the coffee industry will embrace automation in different ways in the near future.
“Automation is about making tasks more efficient, easier to do, and improving workflows. It’s going to make business decisions easier by having all of this information available,” he says. “I don’t think it’s a question of if but when certain technologies will find their place in the industry.”
For more information, visit www.fabscale.com
This article appears in the November/December 2020 edition of Global Coffee Report. Subscribe HERE.
