"Make it Happen: GlobeScope opts for sustainable machine learning

"If we are allowed to create these beautiful, amazing things, let's do it as cleanly as possible."

"If we are allowed to create these beautiful, amazing things, let's do it as cleanly as possible." From this perspective, GlobeScope utilizes Artificial Intelligence to develop smart software for local and regional governments as well as the private sector. We spoke with Erik de Jong, Technical Director of GlobeScope, about their approach, sustainability, and how Leafcloud fits into their operations.

How would you describe GlobeScope's activities?

"Starting from our background, we initiated software development for local and regional governments. Our differentiation in the market comes from combining our knowledge of laws and regulations with experience in environmental services, municipalities, and provinces. Most of our tools are designed to support environmental services in executing their tasks. By utilizing our Smart Privacy Analyzer and Smart Document Analyzer, governments enrich their data and enhance information quality. For instance, we assist in digitizing and archiving diverse paper documents, after which our tools scan the content for useful characteristics for our clients. With our intelligent software, we automatically verify whether the database information aligns with reality. Through our AI components and continuous software training, our tools become more capable, faster, and more reliable. This allows our clients in the public sector to anticipate regulatory changes, conduct inspections, and enforce regulations more efficiently, based on increasingly reliable information. Our smart software is now also employed by businesses, such as the food sector, to extract specific information from documents."

"You don't want to be a polluter when creating these kinds of products."
Erik de Jong, Technical Director GlobeScope

How do you use machine learning in software development?

"To automatically verify, supplement, or remove certain information, the software needs to recognize this information itself. That's where we employ machine learning, utilizing Leafcloud's GPU power to train our models. These are highly intensive processes. Through training, we enhance reliability percentages, enabling us to recognize more document types and other information elements swiftly and accurately. We observe that when digital processes align better, it significantly saves time and costs for our clients. By cleverly applying artificial intelligence and machine learning, we assist governments and businesses in executing their processes more efficiently, such as targeted enforcement and better compliance with the Archive Act, FOI requests, and the upcoming Open Government Act."

What role does sustainability play in software development?

"First and foremost, we aim to create fantastic, appealing software. That's our primary focus and what truly excites us. The next step is: if we have the opportunity to create these fantastic things, let's do it in the cleanest way possible. Everyone is responsible for their environment, so if something can be made greener, we explore that option. Additionally, governments and increasingly commercial enterprises demand sustainable and circular production. As we are a growing company, it means we'll train more models, utilize more computing power, and have higher energy costs. Hence, we seek a provider that operates cleanly or cleaner. We don't want to be polluters. I also consider our social impact; reducing the paper stack with smart software, for example. After extracting data from millions of documents, a significant portion of that archive becomes unnecessary. This paper can be recycled and no longer takes up space."

"I would ideally like to run the entire infrastructure on a green solution eventually. That way, you truly feel like you're contributing."
Erik de Jong, Technical Director GlobeScope

What was it like to start with Leafcloud?

"Leafcloud aligns well with our existing services. I've worked with AWS, which is complex in terms of interface due to its many options. However, precisely the things I want to accomplish there, I can also do with you. Currently, we have a fairly straightforward setup with Leafcloud, not running Apache or similar elements. We spent about two hours together, and then I invested another two hours on my own for connections, and it was up and running. Leafcloud's service aligns with the rest of the market, making it straightforward to use. Honestly, I haven't needed much support. The only challenge we're facing now is the maximum capacity of our GPUs. But there are new, faster cards available, so that should be resolved."

"Meanwhile, the fast NVIDIA Tesla V100 & Tensor A30 cards are available and in use. Check them out here."

Finally, how do you envision the future of GlobeScope and the collaboration with Leafcloud?

"I would ideally like to eventually run the entire infrastructure on green energy. Then, GlobeScope will significantly contribute. I also see where our company is heading and where we aim to be nationwide. This implies continuous growth and ongoing processes. In this scenario, the alignment with Leafcloud only improves; we could analyze thousands of documents every night and ensure people enjoy warm showers in the morning, if technically feasible. If it's achievable, we'll gradually migrate everything to your platform."