Real IDP Use Cases Across Retail, Healthcare & Legal

Part 1 — Intro & Welcome (00:00-02:06)

[00:00:05] William McInnes:
Biggity biggity. Boogity boogity boogity boogity

[00:00:09] William McInnes:
Well, hello again and welcome back to the Enterprise Content Show.
I'm Will, still your host and still the Captain of Content Chaos.

Today, we're turning the heat up. It’s part one of a two-part series about what intelligent document processing is and where it’s already delivering value.

Now, our magicians will make this link happen somewhere. If you click that, you can watch part one.

But part two is getting more specific. We’re diving into industries. We’ll look at retail, healthcare, and legal today.

So, whether you have a legal department or you're in the legal business, we’ve got some really interesting examples. I also love our retail and healthcare examples in this episode.

Who's providing the real insight? It’s RJ, co-founder of Klippa, an engine builder, a document whisperer, and an all-round intelligent document processing evangelist. Lovely human, very smart.

RJ will take us through some of those practical use cases, exploring what happens when you stop sampling documents and start checking everything. What’s the power of the machine to really drive compliance at scale? And what does it look like when machines and humans work together for the best of both?

What does the human-in-the-loop process look like? And what happens when IDP becomes part of the platform, not just another tool, but when intelligent document processing works for you everywhere?

So RJ, welcome back. Let’s get into it.

---

Part 2 — Retail (02:06-04:34)

[00:02:13] William McInnes:
Tell us a story from the world of retail, healthcare, or pick another vertical. What have you seen that's genuinely cool, something you helped make happen

[00:02:13] RJ Verheggen:
There are many. Retail and healthcare both have really cool cases. So, in the U.S., we have a big television manufacturer. Here’s the background: when you’re a TV manufacturer, you make deals with companies like Best Buy and Walmart. They might pay $10 million a year to get their TVs promoted in a certain spot on the shelf.

They’ll say, “For all the stores you have, put the TV here, but don’t put it next to that brand because we don’t want to be compared.” So there’s this behind-the-scenes work. But how do you check if the TV is on display where it’s supposed to be? How do you make sure it’s not on the other side of the store in all these locations?

So, they hired people to visit stores once a quarter, make a physical layout, and write down where everything is.

[00:03:23] RJ Verheggen:
It was super inefficient because people were spending two hours per store creating a floor plan of where everything was.

They came to us with this problem, and we helped them build an app that integrates with Salesforce. Instead of making a physical map and uploading it, the person walks around with the app and snaps pictures of price tags. We can then see exactly where everything is and create a virtual floor map.

[00:04:24] William McInnes:
That’s incredible. They can now do ten visits instead of two. Amazing.

---

Part 3 — Healthcare (04:34-08:15)

[00:04:34] RJ Verheggen:
Exactly. Not sure how much time we have, but healthcare.

[00:04:38] William McInnes:
We love the stories. Keep rolling. Let’s go.

[00:04:41] RJ Verheggen:
I can talk for a few hours, so I’m not sure how long this…

[00:04:45] William McInnes:
We’ll go and get a cup of tea and come back, and you can keep going.

[00:04:49] William McInnes:
Three hours later.

[00:05:01] RJ Verheggen:
You’re triggering with healthcare because I think that’s one of the coolest industries I like to work in, because you really see the impact on healthcare. And if we gain efficiency there, we actually have more people on the job, right? And that’s what you want to get.

Now, hopefully you're not often in a hospital, but what you see there is so much admin burden for all these people working there. And that's so sad to see because they shouldn't be behind the computer logging stuff. They should be talking to the patients.

And anything we can do there is, for me, a personal motivation, more than only a business motivation.

[00:05:44] RJ Verheggen:
So, beautiful cases we've done there where there is, let's say, blood sampling stations where people arrive with what we call like a prescription from a GP, and they need to get their urine or blood tested, and everything is instructed by the doctor.

The doctor just writes down, "Vitamin D? Check. Blood." And then whatever. Cirrhosis in urine, just in their own format.

And they always start the blood sampling stations, and this is impossible to do for computer because one, it's handwritten and two, one doctor can write vitamin D and the other can say this is VD. And we're implied to know what it is. And the people working there, they know what it means because they've done this for a long time.

[00:06:32] RJ Verheggen:
But we actually managed to crack that code now with the modern day capabilities of large language models and just machine learning training in general, that any prescription that comes in, we can understand who's the patient, who's the doctor, what's the date, is it legit, not fraudulent? But also what are the exams that need to be performed?

And so we can already do 60%. We go straight through. So we know that it's good and nobody has to look at it.

[00:07:06] RJ Verheggen:
So at a given blood sampling station, normally there would be three people at the front desk just processing these prescriptions, saying to people, "You can now go into the waiting room."

And the same thing which apply to that cargo case before, now people book a meeting, they get requested to upload the prescription, and when they arrive, they'll know that everything is checked. They can sit in the waiting room, and they can go straight through the sampling room rather than going to that front desk first.

[00:07:39] RJ Verheggen:
And now they went on average from three people to average one person.

And the other two, one on average became like a blood sampler. So more efficiency. And the other is often the hostess. So hey, welcome. How are you feeling? Welcome to your blood sampling station. You want a coffee?

Which is a way better user experience. So that's another driver for efficiency.

[00:08:15] William McInnes:
So, let's go back to the start. So we've got massive efficiency gain. We've got massive scalability potential unlocked. We've got compliance increased and we've got enhanced experience. That's a fantastic example. But more importantly any machine that can understand what the heck doctors scribble when they write down on a piece of paper is a genius machine.

---

Part 4 — Legal (08:15-12:15)

[00:08:21] RJ Verheggen:
Yeah, there’s a lot to choose from. We do a lot of work in gaming and gambling, which involves compliance, but it’s not the most exciting. But one case I like to mention is with cruise lines. They have to onboard thousands of people daily, which involves lots of identity checks. We help automate that. We also work with deceased administration, where companies handle the paperwork for the deceased, and we’ve automated that as well.

But one of my favorites is in the legal space. There’s a lot of paperwork and inefficiency in legal work, especially with case administration. But one example I wouldn’t call weird, but unexpected, is working on cartel damage claims.

[00:10:00] RJ Verheggen:
So, in Germany, there was a cartel fined for price fixing in the pesticide market. They made illegal pricing arrangements that harmed farmers. The farmers, incentivized by a law firm, came together to get their money back. The law firm collected ten years’ worth of invoices to prove the farmers overpaid.

The challenge was sorting through 100 million invoices to extract pesticide prices and calculate the overpayment. They realized if one invoice was off, the whole case could fall apart. So we helped automate the process, ensuring that every invoice was processed accurately.

[00:11:22] William McInnes:
That’s an amazing example. It makes me think: with IDP, we really do have the receipts, right? No archive is too vast, and no complex bag of unstructured information is too much for this technology. It’s changing fields like legislation. The ability to process all this paperwork efficiently is a huge game-changer

[00:12:15] RJ Verheggen:
Yeah, it’s about looking at where you have people doing repetitive tasks. With IDP, you can make their work more valuable. You can also clean up archives and check for things like personal data, then automatically redact or delete it.

---

Part 5 — Looking to the Future (12:15-16:19)

[00:13:13] William McInnes:
As we look to the future, RJ, what’s exciting you most about intelligent document processing? What’s keeping you and your team at Klippa interested in the next step?

[00:13:51] RJ Verheggen:
I’m a techie, so I’m thinking about the next big thing. What we've used so far are mostly generic models—models that can process invoices, for example, but they have limitations. You can never get to 100% because there’s always variation. But with Doxis AI.dp, we can now create custom models for specific use cases and data. It’s so much easier now to train a model tailored to your data, so we can apply a custom solution instead of a generic one that only works 90% of the time. This new approach can do things that humans can’t, like fraud detection, at scale.

[00:14:59] William McInnes:
That’s wild! So, you’re saying there are truly new frontiers ahead, and we’re not just talking about minor improvements, but huge advancements

[00:15:43] RJ Verheggen:
Exactly. A few years ago, we thought accounts payable was the obvious case—any business has invoices, so a model can handle it. But now, with industries and use cases becoming more specialized, we can design models that work for your unique needs, no matter how complex.

[00:16:16] William McInnes:
Very cool. Alright, RJ, that was fantastic. Thank you. We’ve talked about scale, compliance, human-in-the-loop models, and how IDP moves from theory into practice. At Doxis, we’re helping you run operations smarter, leaner, and more efficiently. It’s not about hype—it’s about real-world examples from RJ and our customers.

To everyone watching or listening, if part two sparked any ideas or questions, like it, share it, and subscribe. Ask us questions—we’d love to hear them. If you haven’t watched part one yet, check it out. RJ, thank you again. You and your team are legends. And to everyone else, we’ll see you in the next episode of The Enterprise Content Show.