How does riskmethods’ risk identification really work? An interview with Director of Risk Intelligence Kasia Katrycz
by Ali Jawin
Buyers Meeting Point’s Kelly Barner interviews riskmethods’ Director of Risk Intelligence Kasia Katrycz on how her team combines artificial intelligence and machine learning with human verification to provide actionable intelligence on potential supply chain disruptions.
Below is a condensed version of their interview.
Kelly: Tell us how you work with, and alongside, the technology available to help companies stay on top of constantly shifting risk conditions.
Kasia: We live in the era of information with more content being created every day. Our task is to find the right information at the right time. We navigate the millions of online data sources with six filters. The first filter is to monitor only what is important for our customers – their specific suppliers, customers, transportation hubs or anything they tell us is relevant. The second filter takes the information flagged from the first filter and runs it through search engines like Google. The third filter is where AI comes into play to help us master big data searches. Only AI can analyse vast amounts of data and narrow those results down. The fourth filter is what we call “machine learning algorithms and logics” which help us to weigh pieces of information against each other to define what information we should look at first. The fifth filter is the human component where the analyst verifies that the AI and machine learning has in fact flagged accurate information (providing feedback to the machine when necessary). Finally, the analyst enriches the alert with information found in the context, defining a radius around the epicentre of the disruption so all potentially affected risk objects within that radius are apparent to our customers. Our goal is to not be another newsletter that nobody looks at. Our task is to deliver actionable information.
Kelly: How do you establish levels of how serious something has to be before you issue an alert?
Kasia: riskmethods has a scorecard with about 80 indicators. Most of them are in this daily monitoring so we already have a scope of what risks warrant an alert. However when customers receive an alert from us it doesn’t necessarily mean there is a crisis, we also send signals for issues that may unfold that are worth paying attention to (such as plans to shut down a plant).
Kelly: Even once you do find something that appears relevant there is always a quality check because you are talking about information from the internet. What is the quality assurance process to ensure that the information is reliable before the alert gets triggered?
Kasia: That’s exactly what my team is here for and why human analysis is so important. The technology and our analysts are working around the clock to ensure alerts are reviewed for relevance and enriched as much as possible. It’s very important that our customers trust that if they receive an alert from riskmethods, it is important.
Kelly: Can you give me an example of alerts that you have recently passed through?
Kasia: An interesting example occurred during hurricane Matthew. I was in the US during the hurricane so of course I heard all about it, as did all of our US customers, but that wasn’t the only hurricane that week or the only hurricane to affect our customers. There was another hurricane in South Korea at the same time, but if I was only relying on the news in front of me, I would have never of known. The hurricane in South Korea illustrates that the information we get from TV and our national news sources is only the tip of the iceberg. 90% of news isn’t covered in mainstream media channels because it isn’t eye-catching or relevant for most people, but that’s where most of supply chain risk news is covered. That’s why we listen to everything and then determine if it’s relevant for our customer.
To hear more about the interaction between human beings and machines that make it possible to isolate relevant risk indicators for our customers, listen to the rest of the interview here:
Learn more about how riskmethods’ Risk Intelligence works here.