How are businesses managing the growing complexity of unstructured data? This episode explores how Intelligent Data Processing (IDP) is reshaping operations, increasing efficiency, and supporting smarter decision-making in industries like banking, insurance, and healthcare.
HOW INTELLIGENT DATA PROCESSING IS ROCKING BUSINESS GROWTH
PROTAGONISTS
TRANSCRIPTION
HOST Introduction:
Welcome to The Power of Possibility. Today’s episode is all about Intelligent Data Processing, or IDP—a transformative approach that’s helping businesses improve efficiency, reduce processing times, and elevate customer satisfaction.
According to IDC, by 2025, 80% of the world’s data will be unstructured, presenting huge challenges for businesses. Enterprises will need to rethink their strategies for data management in order to take advantage of their full potential. IDP has been already rising to meet these challenges. It’s cutting manual workloads, automating processes, and even driving measurable growth.
We’ll hear from our expert guests today about real-world applications of IDP and its role in industries ranging from banking to insurance. So, let’s dive in and see how IDP is reshaping the future of business.
HOST: So, Christian, to kick things off, could you explain what Intelligent Document Processing (IDP) is and how it fits into the broader context of data processing?
Christian Schierjott: Yes, I mean, IDP or intelligent document processing is all about digitization of data and documents. Especially, I mean, what we see today, physical documents, but also digital documents that come in, in the inbound processing of companies. So actually, what we're talking about is mainly the classification, the extraction, and the processing of especially unstructured and semi-structured data. The "I" for intelligent comes really from the combination of ordinary technologies like an OCR, what we know for quite a while, but now in the combination with cognitive elements, for example, with natural language processing, or even now machine learning or generative AI, it really becomes a new dimension of document processing. I mean, I would say that especially the structuring of this data is a key functionality of the IDP technology. And it's really focusing on analyzing and interpreting data flows and by this facilitating the decision-making process of companies, which is very important, becomes even more important in the future.
"What they were able to do is now in minutes—sometimes 15 minutes—fully adjudicate that life insurance claim while they have the person either online or on the phone. If anything's missing, it notifies right away so that they don’t have to call a person back and remind them about this horrible situation." — Adam Field
HOST: Adam, from your perspective, how is the landscape of document processing evolving, and what role do you foresee for intelligent automation in the next five years?
Adam Field: Sure, sure. Well, I think it's really interesting because intelligent document processing or IDP has been around for a very long time. But just in the last few years, I think the first thing we need to do is define exactly what it is. Because I think intelligent document processing generally, the definition has expanded. In fact, the term might not even make sense anymore because it's about more than just the documents.
Really what organizations want to do is make sense of all the information inside of their organization and automate what can be automated, regardless if the document is at the center and starts that automation or is just an asset as part of that overall process that they're trying to automate. So I think IDP used to be a lot about just maybe taking data off of forms and doing data entry into another system.
And then moving that information somewhere else and processing it somewhere else. IDP has really expanded where now I think it involves case management and workflow and RPA and a user experience across different channels and almost moving, also moving beyond the document, meaning being able to process information regardless of where it lives inside or outside of an organization.
HOST: Christian, many businesses see IDP as part of a larger trend in advanced data processing. How has data processing evolved over the years to meet the growing demands of modern businesses?
Christian Schierjott: I think that's a good and valid question, right? I mean, it goes back a couple of decades. I remember when I started my career in a bank and they had all these computers, right? Which is, I would say, the early stages of data processing from a technology standpoint. But it really took off with IBM Watson nearly 10 years ago over to RPA—robotic process automation—. I mean, we as SPS had a couple of bots implemented for our clients to help them automate data processing. Now, if you look at where we are standing now with really this natural language processing and generative AI technology, this evolution cycle gets shorter and shorter over time and enables us to do more and more with this technology.
So looking from automating simple manual tasks, typing data from one system into another, and seeing now what you can do with modern technology, not only automating very simple manual tasks, but even highly complex tasks like decision-taking is a huge step in the evolution. And I'm pretty sure that in the near future, there will come more and more of more complex tasks that can be automated in the field of data processing.
HOST: What kinds of inefficiencies or redundant tasks do businesses typically face, and how does intelligent data processing address them?
Christian Schierjott: I think that when we look into companies, there is a lot of manual data entries, be it that I have to take our data from forms into my system, reading lengthy emails or letters and compiling data out of it, or handovers of processes in document handling, but also siloed data, which is slowing down really the decision-making process.
I really see a lot of tasks which are redundant or highly inefficient. And that's what we're focusing mostly at SPS on is really automating this repetitive task, reducing the handling errors, and by that also the cost. But also due to scaling easily to handle increasing document volumes, which is still the case when you look especially into insurance companies, for example, in the health sector. It's not really seen yet that we have a massively declining volume of physical documents or electronic documents that are scanned and come in to our clients.
HOST: Christian, you’ve mentioned the concept of “unstructured data,” which is becoming increasingly common and presents challenges for businesses. Could you explain a bit what unstructured data is, and why managing this 'data chaos' is so important for companies?
Christian Schierjott: Unstructured data is something that everybody knows. If you just write a letter to your telco provider, for example, if you write emails to them, usually every form has a somehow free text field in the form where you can put in a comment or a handwritten request. That is all the fields that are very difficult to extract and to read out information because everybody can just type in different kinds of information.
But on the other hand, especially those free text fields or emails or letters contain a lot of useful or needed information that I have to extract from those documents. And that makes it especially hard and challenging to work with those free text and unstructured information.
HOST: Adam, expanding on this, what unique challenges do industries like BFSI and healthcare face in managing unstructured data, and how does Tungsten address them?
Adam Field: Sure. Well, unstructured data itself is an interesting term because when I talk to business people, of course, IT people fully understand what that means. But when I talk to business people, sometimes we spend time defining what is that. And so if I were to say, unstructured data, typically it's been information or data on a document that isn't formatted in any particular way, just free-form text like a paragraph.
But that definition itself, once again, is expanding because if you think about all the information that an organization has about a customer, those could be transcripts from audio files, video files, social media posts, other public information like news stories if you have to do background checks. And no one knows what kind of format that's going to be in.
And then that helps, I think, set the table for, OK, what is unstructured data? And now that we define it, you know, I think what's really unique about what the Tungsten platform provides is on one common platform, you can do that intelligent document processing that we talked about a moment ago. You can do full process and workflow automation. And then it has many AI capabilities like document AI, generative AI, decisioning AI, knowledge bases, AI agents built right into it. So, we just look at unstructured data alongside any other kind of information, use the right technology to elevate it so that humans and machines can make sense of it, and then try to automate what we can inside of your organization and bring humans in the loop as necessary.
HOST: From your point of view, Adam, what are the main problems that companies face when they come to you? What challenges are they dealing with?
Adam Field: Well, it's going to vary a little bit depending on the industry that the customers are in. But generally, what they're telling me is the amount of information coming into the organization is growing faster than they can handle it, faster than they can implement IT projects. Their budgets have already been cut. They're looking for strategic vendors that can do more than maybe what they asked them to do in the past. And essentially, they have to do more with less.
Customer expectations have grown. And so they need to get the right answers to the customers. They need to process things in a very efficient manner. But the challenge across all of that is they have to do it ethically. They have to do it compliantly. And so there's some trust that goes into the technologies that they choose to do all of this.
It's a big challenge inside of organizations balancing great customer service and automation and now adding the use of AI and generative AI. Can they trust it? How should they use it? Where should they use it? It’s a really big challenge.
HOST: Christian, shifting to industry specifics, industries like BFSI and healthcare are heavily data-driven. What’s the current state of data processing in these sectors, and what unique challenges do they face?
Christian Schierjott: Yeah, that's a very good question. I think that first of all, which is for the entire BFSI sector in common, is that they are heavily regulated, especially when it comes to customer data. I mean, we know that from the banking industry, with banking secrecy, we know that from the health sector, with patient data, for example. So this is always a critical point when you deal with those industries.
Another challenge is, and I would say that is also very common for the entire industry, is really staying on top of the huge amount of data the companies collect throughout their customer interactions and through their operational processes. So they have a huge amount of different data they have to deal with, they have to interpret and take business decisions on.
And then you have more challenges which are specific to certain industries, like the banking industry managing fraud detection and compliance reporting in the area. Healthcare, as I said, it's if you deal with insurance claims with patients' data, very sensitive data. And I think that's quite a challenge, especially when you want to partner with external service providers to give out such highly confident and sensitive data.
HOST: Christian, could you share an example where intelligent data processing turned a data challenge into an opportunity, perhaps improving customer satisfaction or operational efficiency? How does SPS provide solutions in this area?
Christian Schierjott: We have a couple of those examples at SPS. I think the latest one is our SPS GPT platform, which we developed back in December 2023, when this entire generative AI hype came up.
We developed a loan application process for one of our clients, where we really shortened the application time for a private loan, making it much easier for our clients to handle those loan applications more efficiently. So overall, you have a better quality of the entire process. The process and the loan approval are given much faster to the client. And so is the payout to the client in the end.
That really is a good example of how we serve our client—in this case, the bank—to save costs on the one hand, to increase the quality of the process, but also on the other side, to help the end customer get the loan in a much faster time.
HOST: Adam, could you share a specific example where Tungsten Automation helped a business overcome a significant digital workflow automation challenge, leading to measurable growth or efficiency gains?
Adam Field: A couple of examples come to mind. One is a very large bank, and if I said the name, you would definitely know who they were. They're doing their full mortgage processing in Tungsten.
So this bank operates in a very large country with a large population in a very competitive mortgage environment. They’ve cut processing times for mortgages from days, sometimes weeks, often down to minutes.
And what's interesting, I think, about this use case is the information that they're using to process these mortgages can come from documents, but also information from other systems. Still, their process involves often customers or prospective customers walking into a branch and handing some income documents and other paperwork to the person at the branch.
So they're scanning them right in the machines at the branch. Regardless of where the information comes in, there's a centralized process in Tungsten that takes care of that. And Miguel, to my point earlier about sometimes it starts from a document and sometimes it doesn't, you don't know where, especially today, your customer is going to initiate this work. Maybe they walk into a branch, maybe they go to your website, maybe they walk into a broker. It doesn't matter. It doesn't matter where it starts, what type of document it is.
We're processing all of that information, integrating with the other systems they have in place, and then automating the decisions because the quicker you can do that, then you're likely to lock in that customer. But if a human does need to be brought in to make a decision, we're doing that quite seamlessly. So it can always be tracked by the customer and by the bank.
So sometimes days or weeks down to minutes. There’s also a really big insurer. I was talking to the head of their digitization operations the other day. And she told me a really interesting story that, sometimes, we get so wrapped up in our day-to-day work and the technology, and you forget about sometimes the human story behind some of this.
And they're using Tungsten to automate their life insurance claims process. Now you can imagine when a loved one dies, someone has to pick up the phone and call the insurance company to make a claim on their life insurance. And that can be a very difficult time.
What they were having to do is these people that were processing their claims, while they're trying to be empathetic, they also had to tell these people that it might take days to get this stuff processed because they needed paperwork. And if the paperwork’s not right, they had to pick up the phone and call back.
What they were able to do is now in minutes—sometimes 15 minutes—fully adjudicate that life insurance claim while they have the person either online or on the phone. If anything's missing, it notifies right away so that they don’t have to call a person back and remind them about this horrible situation.
And most importantly, they’re able to get the customer their money sometimes within minutes.
“I think what we see today with the evolution of large language models is just a glimpse of what is actually possible in the future". - Christian Schierjott
HOST: Christian, how does data processing impact customers’ everyday experiences, particularly in sectors like banking and healthcare?
Christian Schierjott: I think with every interaction you have with a company, be it in the healthcare sector, banking, or insurance, there's always reliance on data.
For example, with an insurance claim, you send over data to your insurance company, be it in an email, form, or call. Nowadays, also calls are recorded, and you can take out the data. Speech-to-text is a keyword here.
In the banking area, if you want to apply for a loan, it’s always about the amount of data and how you provide it to the bank. With loans, everyone knows what you need in terms of documentation to get a mortgage or a personal loan.
So, it’s a matter of how you efficiently transmit those documents and data to your bank, so that they can efficiently process the data to get quick results at the end.
HOST: Christian, adopting new data processing solutions can bring risks. What are the most common ones, and how can businesses safeguard themselves?
Christian Schierjott: I think there are two sides of it when we talk about the risk. First of all, it's a cultural question in organizations. People must be ready really to think processes differently, especially when you have very new technology like now with large language models where you have totally different possibilities to build an operational process. So organizational resistance is one significant risk.
The other thing is obviously the technical obstacles you have where we have the risk of data quality, misleading interpretation of data, and data security and privacy breaches. This is a big question, especially if you have pre-trained or constantly trained systems.
And therefore, I at least have the hope that with the European AI Act, we get some clarity on that when we implement such solutions that we're really on the safe side here from a compliance perspective.
HOST: Adam, if you had to predict, what groundbreaking advancements do you think will redefine IDP by the end of this decade?
Adam Field: By the end of the decade? Well, who knows what's even going to be by the end of the year as fast as things are moving. But if I had to say in the next five years or so, I'd focus on a few things. One is democratization. And by that, I mean the democratization of both technology and insights into information.
With generative AI, we rolled out here at Tungsten Automation in early 2024, a Tungsten co-pilot based on generative AI and large language models. It helped democratize a few things. One is development time. Low code has been around for a long time, and that's slowly going to be replaced by just prompting. So we're going to build software using natural language, using our voice. You can just type in any language, and it will build workflows, screens, and business rules based on context. Similarly, you can build your document extraction models. Now, instead of scripting or drag-and-drop, you just type prompts and it determines what to extract from a document regardless of format.
So training time goes down. That’s the democratization of technology. And on the other side, more of a business-focused runtime angle, there’s the democratization of information insights. We also rolled out another Tungsten co-pilot for insights which gathers information from documents related to a case that’s been processed and information from elsewhere in the organization and just with some prompting, within seconds, I can get insights into that data that would have taken humans hours to process before.
So if you ask me over the next five years, I think that just makes people more productive. But with that comes a little bit of fear, right? So people are fearful. They hear about AI, they hear about gen AI, and they're fearful that it's going to replace them.
I heard a great quote once: “AI will not replace you. Someone who knows how to use AI better than you will.” And I think, in five years, it’s not like we’re going to be sitting still doing nothing while computers do everything. Jobs have continued to increase even as tech innovation has increased over the last 40 years. It’s going to make us more productive. It’s made me more productive. I started coding again after 20 years. It’s going to take mundane work. People will realize that they're not going to be replaced, that these things are assistance to help them do their jobs better and make them more productive.
HOST: Christian, if you could peer into the future, what do you think data processing will look like in five years? How will it transform businesses and industries?
Christian Schierjott: I think what we see today with the evolution of large language models is just a glimpse of what is actually possible in the future. I think that we will see much more efficient data handling, data management processes, especially for unstructured data.
When I see what we are able to do now already in our service processes in the telco and banking industries, I think that the handling and management of data in our day-to-day processes will fundamentally change. Also, the skills we need for data management will change.
We will see more skills in handling and maintaining new technology than we have today with agents sitting at desks serving customers or dealing with applications for loans, onboardings, or claims.
HOST: Finally, Christian, what advice would you give to companies looking to invest in or optimize their data processing strategies for growth and efficiency?
Christian Schierjott: Looking back on the last couple of years at SPS, we gained a lot of experience in developing new ways of data handling, data management or document processing.
And I think the key takeaways, at least for me or for us as a company, is really to start small, focus on high-impact use cases, demonstrate quick value and not try to change the world at once. Another thing is really to prioritize security. Choose providers with strong compliance capabilities, which are proven in the market.
And last but not least, collaborate with experts who really know their stuff, who have experience for a long time and can demonstrate it also. At SPS, we leverage the experience to design tailor-made solutions and that over the last two decades. So that would be my advisory: to look out for partners who demonstrated experience in the market for a long time.
RELATED CONTENT
Revolutionizing Document Processing Operations
In today's rapidly evolving business landscape, companies are constantly searching for new ways to streamline operations and improve productivity.