00:06 Jennifer Poma: I'm Jennifer Poma. I'm our Director of Marketing here at UDig and I'm here with our CTO, Josh Bartels and our EVP of Consulting, Reid Colson. And they're gonna share with us some insights based on their most recent thought leadership piece around the power of document intelligence.

How does unlocking unstructured data through document intelligence change the game for a business?

00:28 Reid Colson: You know, I think about it and most of the data that an organization has is not in their data warehouse and it's not structured and it remains largely untapped. A lot of folks call it dark data. And so document intelligence is gonna transform that chaos into actionable insights by automating extraction, classification, and analysis.

And so instead of scanning through hundreds or thousands of documents to find answers, you can simply ask questions and let the AI do that work for you. And so now teams can, you know, supply chain, look for a supply chain bottlenecks. They can find compliance gaps or, you know, they can find customer trends and data they couldn't look at before.

01:08 Josh Bartels: And I would add to that. I think you also are starting to get a unique capability to start to aggregate some of that information across those discrete documents as well. So you're also unlocking capabilities of, you know, summarizing trends that might be locked in documents or tables that have changed over time inside of a document.

So we're also seeing some unique capabilities in that area where you can open up like vast amounts of knowledge, but also kind of do it in a way where you generate aggregates on top of those documents to give even more insight.

01:47 Jennifer Poma: How is document intelligence going to create impact for organizations?

01:52 Josh Bartels: So thinking through potential couple examples of how we've seen it work today and what potentially could be happening in the future, even things that we're potentially doing internally. So one thing that we've done internally using kind of the concept as it relates to document intelligence is opening up all of our pre-sales information to our growth team and allowing them to explore the information about different concepts or approaches that we've taken with clients that gives them better access to, you know, 700 documents that are in a SharePoint site, right? Without having to explore those 700.

And the impact that that really creates is for knowledge workers, they don't have to go find it, right? So the efficiency and time that it takes to do that. And then the other impact that it creates is the content that they can generate from existing content also is beneficial in the sense of when you go to create content, a lot of the problems that you have is that first initial step of generating it.

It's a lot easier to edit it than it is to generate it. So having that corpus of information that you can access, ask questions about, and then generate from really creates a speed and efficiency, both from finding information, but also in the creation of content that is relevant, relevant to the outputs that you're trying to create. And so that's been really beneficial internally.

As our clients have been using it or other areas that we've been exploring, there's definite use cases in different industries for how this potentially works, whether it's somebody who manufactures homes and in those homes, they use the same set of refrigerators, dishwashers, et cetera, in allowing access to the manuals, right? So homeowners can ask questions of the manuals versus having to like look through the manuals and create kind of that experience of how can I get help or support on these items that I have without having to look through 100 and 150 pages of manuals because I'm just trying to figure out why my dishwasher is blinking three times yellow and understand what that means, right? So that on the flip side there, the impact that that's really beneficial is there's two things, right?

One, the customer can find the information without having to call somebody or really look through that manual. And the other impact it potentially makes is by knowing the context of the conversation that that customer is having, if it is something that really requires service, you can now flip that into a service request really quickly, right? And help reduce the friction between the customer and the service center by basically reducing that time in question ask because they've already done some conversation and now you have a lot of the information to help generate that service request and reduce the friction to get support for what you're looking for. So just a couple of use cases, one internal, one with our clients.

05:21 Reid Colson: Yeah, I think what I'd add to that is you have lots of roles where folks are doing reconciliation or compliance and being able to look at the documents that are specifying a contract, like an example might be if you're a large nationwide chain that has janitorial services for all your facilities across the entire nation, you're doing acquisitions, you start to find in certain marketplaces some of these services are available across marketplaces and you can start to compare between services and look for ways to optimize and create value by spending less or getting lower risk in some of the services that you're purchasing. It's an exercise that would take a person a really long time to do and it's a lot of reading and comparison, but you can let the tools interpret the documents and give you those insights really quickly.

We're seeing it with things like invoicing and payments and billing.We're seeing it in things like evaluating red lines in contracts or changes in contracts that didn't get marked appropriately to give you insights without having to read, remember and find the insight yourself.

06:31 Jennifer Poma: So why is now the right time for organizations to invest in this kind of capability?

06:36 Reid Colson: You know, I think now is the right time for organizations to invest because there's a confluence of things that are coming together all at the same time. One, the capabilities of the tools are getting so much better. So what Gen AI can do with, you know, RAG technologies and the ability to interpret what's in a document is getting much more powerful.

It's also getting much more affordable and it's coming at a time when, you know, almost 80% of an organization's data is unstructured. This is according to IDC and they're finding that folks are spending two to four hours a day searching documents for information. So it's a significant productivity lever that folks can use at a time when the capabilities are getting better, becoming less expensive, when we're all pressed for better economic outcomes in our organizations.

07:24 Jennifer Poma: What would you say to a leader who still sees documents as static files instead of strategic assets?

07:31 Josh Bartels: I would say that documents do change over time, right? They're not just static files, right? So documents will get edited or you'll copy documents and create new versions of documents.

But what is truly kind of the value that's inside the documents is really what they should be thinking about. So don't think about the document, think about the content that's inside the document and what, if you looked over a bunch of documents that you had in your organization and you thought about the content that was inside of those, knowing that what questions do you think you could answer that you can't answer today, right? So we've talked about compliance related questions.

We've talked about trending related questions, right? So documents or contracts are often written into, so contracts are often written as documents, right? And they're not discrete data elements, they are a contract.

But asking things like we know in a contract, we know the values that it might cost, we know the time of that contract, right? So, hey, we can say of all the contracts that we have with our different vendors, which contracts are trending 20% higher year over year versus 10% higher year over year, that discrete data is inside the content of the document. So don't think of it as just a discrete file, think of it as what questions could you answer if you're really looking at the content.And the value that could provide your organization.

09:09 Reid Colson: I think Josh nailed it. I think what I might say is if your documents aren't a strategic asset, I'd wonder why you're creating them and storing them in the first place.

The fact that you are creating and storing them tells me that there's something about them that's going to be valuable for you. And without investing in document intelligence, you don't know what you don't know.

09:28 Jennifer Poma: What excites you the most about the future of document intelligence?

09:32 Reid Colson: I'm gonna be a little selfish on this one. Honestly, what excites me the most about it is the ability to grow my knowledge and my personal productivity. So I envision a world where I've got a personal agent that's looking at all of the sources of information that are important to me, summarizing it and creating an audio file that I can listen to on the way into my car, right?

So I can use that 15 or 20 minutes of my commute every day to get more effective, learn something, knock off items off my checklist and not eat into the rest of my day and be able to use that for other things.

10:01 Josh Bartels: I would agree with Reed of, as we move beyond kind of step one in document intelligence about being kind of like a search and create function and starting to move to automation and even agent or agent-based workflows, right? How, what I'm excited about is how do we take those things, you know, Reed's kind of, you know, golden agent idea that he had right there.

There's probably hundreds of more ideas that can kind of go along with that. And I'm excited to see what those potential use cases and opportunities are as we move towards more of the automation that could go along with document intelligence and how we could potentially personalize it, right? So even in Reed's case, right, how do we know what documents are important to Reed so that when we do create the summary, it's not a summary of the 100 documents that came into his inbox over the week.

It's the 20 that are important because I know Reed's preferences or personalization, right? And so it's the additive features that are building on top of each other. And like I turn around tomorrow and I got to read the latest news because somebody created some new feature capability.

It's exciting, but it's also a lot to keep up with, but I'm just excited about the constant changes in the marketplace.

11:28 Jennifer Poma: Why shouldn't I just use whatever tools come with my document repository for document intelligence?

11:34 Josh Bartels: So I'm going to answer this question kind of maybe more in a general or holistic in the sense of when you evaluate capabilities or features of any platform, right? I think you should look at what you are truly trying to achieve and what value are you trying to get. And if your built-in tools don't have the capabilities that you need, you should go out to market and find something that potentially fills those gaps, right?

I would say the good things that partner with existing document repositories are going to be more of the, they know a lot of like your permission model already. So like they really help with the security and access of documents, right? But that could be a two-edged sword, right?

If you work in an industry that is really reliant on that security, then it makes a whole lot of sense. If you work in an industry where sharing documents across everybody is more beneficial, right? That permission model could really restrict you from what you're trying to achieve or what you're trying to do.

So my approach is always kind of looking at what am I trying to achieve? Does that exist in the existing capabilities or what I have? Before I go out and procure a completely different solution when it relates to document intelligence, I might look for something that integrates with my existing repository to give me those capabilities and features.

But if I can't get it that way, then I might go out and look for something that might give me that entirety. And we're seeing kind of, what we're seeing now is a lot of the frontier model or foundational model, right? They're changing their names as we go through this right now.

Organizations like OpenAI, Anthropic, Perplexity is just a wrapper around some of that stuff. But these major platforms that folks are using are starting to even take the concept of document intelligence and bring it to them. So you don't necessarily, you can abstract away from your document repository of SharePoint or Google Docs or Google Drive, right?

And you can connect across those and get document intelligence across multiple different things, right? So then that becomes more of a question of, are these things gonna train their models on my data, right? So now you just have to start to look at the AI components and what the things that you're using in those AI components and understand how they're gonna be used by those organizations, whether or not you're an organization that's really high on security or an organization that's not, or you're just an organization looking for really advanced features or really just more of the, I think if you're looking at traditional tools, they're gonna move slower than some of the newer platforms because those traditional tools have to support a much larger customer base as they grow versus the newer platforms are trying to just release features and grow the amount of customers that are on their platform.

So just a couple of things to think through as you're exploring what exists in the ecosystem.

14:46 Reid Colson: Yeah, I think that last point that Josh made is super important. The ability to advance the capability for one company is not gonna be as strong as being able to use any company.

So if you get locked into the one tool, you do probably have some limitations in your capabilities over time. I think the other angle I would think about is whether or not all of my data is in that document repository or all of my documents are in that repository. And if I only wanna ask questions that look at the documents.

So if I want to combine a document search with my structured data, that's gonna be a lot harder to do through the tool that is sitting in that document repository. And it's likely that some of the questions that you wanna ask are gonna span your documents and your structured data. So having the abstraction to be able to bring different types of document intelligence tools to bear on your given problems is gonna give you a lot more of that flexibility.