The Gold Mine of Untapped, Unstructured Data Hiding in Decades of Enterprise Paper Records

09.22.2026

As businesses race to scale AI, intelligent intake—the process of identifying, validating, contextualizing, and routing business information—is bringing new attention to an overlooked question: what happens to data the moment it enters your organization? Imagine that you work for a large insurance company that processes thousands of claims every week. At any given moment, an abundance of information is entering the business—whether it be through online forms, email attachments, broker submissions, or traditional mail—all of which need to be identified, validated, matched to the right customer, and routed to the right place before a claim can move forward. You might assume most of that happens automatically. In reality, employees inside large organizations are often still opening attachments, re-keying details, cross-referencing systems and deciding where information belongs. It’s an invisible yet significant back-office problem. Then, enterprise AI entered the picture. Information intake has always created inefficiency, but now, AI is seriously raising the stakes. If your business is using AI to analyze information, recommend decisions, and automate more complex workflows (or plans to in the future), the quality and context of the information feeding those systems becomes much more consequential. For that AI investment to pay off, the right information must reach those systems in a form they can understand and trust. The AI Race Is Running Into A Data Problem For the last few years, much of the enterprise AI conversation has centred on models, copilots and (increasingly) agents. But as organizations try to move AI from impressive demos into real business processes, something much less exciting is getting in the way: their data. In a 2026 study of executives at 2,000 companies, Accenture found that 64% said their organizations had moved beyond isolated AI pilots or begun coordinated enterprise-wide adoption. Yet just 7% had progressed far enough in building the data capabilities considered necessary to support advanced AI at scale. McKinsey has identified a similar constraint. As organizations push AI pilots toward scale, data readiness becomes a limiting factor, particularly when businesses must connect structured data with the vast amount of information held in documents, emails, and other unstructured formats. In other words, buying or building the AI is only part of the problem. You also have to give it something worth working with. And that raises a question that gets far less attention: What happens to all the new information entering your business tomorrow? The invoices. Applications. Contracts. Customer emails. Forms. Claims. PDFs. Images. API calls. Physical mail. Sure, cleaning up the information you already have stored in file folders, bankers boxes, and digital systems is a challenge. But making sure you aren't creating tomorrow's data problem with every new piece of information that arrives is another. The Front Door To Your Business Is Still Surprisingly Manual Inside large organizations, information rarely enters through one neat channel. It arrives everywhere. This is why a process called intelligent intake is taking centre stage. Matt Christensen, Director of Digital Transformation and Innovation at WCD, describes intelligent intake as governing those different entry points and turning incoming information into valuable context the business can use. The goal with intelligent intake is to make all forms of information coming into a business usable, whether that means routing it intelligently, starting a workflow, or giving AI the context it needs." Without intelligent intake processes in place, Christensen has firsthand seen how quickly complexity can build inside large businesses. In one government organization, employees were responsible for sorting incoming information across more than 100 different document types, each with its own place in an extensive filing structure. Knowing exactly where something belonged often depended on experience and institutional knowledge, which became especially challenging for smaller offices with fewer people to rely on. Intelligent intake can take much of that guesswork out of the process by identifying what a document is, extracting the relevant information, and applying the right classification and context as it enters the organization. That matters even more as businesses begin connecting AI to their enterprise information, because a document that is misclassified, poorly labelled or difficult to find can affect what information an AI system finds, how it interprets it and, ultimately, the answer or action that comes next. Why Your Enterprise AI Needs Contextual Clarity Giving AI access to your company's information is one thing, but giving it enough context to understand that information is another. Christensen points to the oil and gas industry to paint the picture for us: a general-purpose AI model might know a tremendous amount about chemical engineering, but it doesn't know your company's standard operating procedures, the history of a particular well site, its pressure readings, or what happened during its last maintenance cycle. That knowledge is specific to your business, and it often lives across documents, systems, and the people who know how to interpret them. Context is king with AI. If you don’t control the context your AI has access to, you can’t control the quality of what you get back. Intelligent intake becomes part of that context control.” Think about something as ordinary as a customer complaint. Knowing that an incoming email is a complaint is useful, but knowing that it comes from a major customer, references a recurring issue, and contains language suggesting possible legal action signals to run this particular complaint up the chain of command, immediately. When incoming information can be understood with a contextual layer applied, the system can make a much more informed decision about how to move or escalate it to correct people or systems. Intelligent intake gives us the ability to route information based on context, not just document type. You can understand the urgency of what’s happening and bring in information from other systems to determine what should happen next.” — David VanDerEems, Senior Consultant, Enterprise Transformation That's where intelligent intake starts to look very different from even traditional automation. The goal is multi-purpose: to get information into the business faster, and to make sure it arrives with enough context to be useful once it's there. Taking Business Data From AI-Ready to Agent-Ready The flip side of giving AI access to more of your business is that bad information can travel further, too. VanDerEems comes back to one of technology’s oldest principles: “garbage in, garbage out.” If you’re using AI to help make important business decisions, governing the information going into it becomes critical. The higher the stakes of the decision, the more important it is that you can trust the data behind it.” This is especially important as AI moves from answering questions to taking action. It’s no longer just “AI-ready data” but also “agent-ready data” that organizations need to consider. AI agents increasingly use enterprise information to reason, make recommendations and initiate workflows. And because of this, the progression of enterprise information management is starting to look something like this: Intelligent Scanning: Turn historical paper documents and other physical records into digital information that can be searched, accessed and used across the business. With an added layer of data extraction, that can also mean capturing the information inside those documents so it can be used by other systems and your enterprise LLMs.Intelligent intake: Understand all forms of data as they enter the organization's many “front doors” for information (like the mailroom, customer service inbox, or support call transcripts). That means identifying what it is, extracting and validating key information, adding the right context, and routing it based on what the business needs to happen next.Automation: Put that digital information to work by moving it through repeatable processes with less manual intervention, whether that means routing an invoice for approval, updating a customer record, or triggering the next step in a workflow.AI & Analytics: Give AI systems the ability to use that trusted, contextualized information to reason across business data, make recommendations, and increasingly take action with agentic workflows, from resolving routine requests to initiating actions across multiple systems. Each step expands what a business can do with its information, and raises the cost of getting that information wrong at the beginning. Weak intake does not stay at the front door; it compounds through automation, analytics, and agentic workflows. Your AI Strategy Has A Front Door When leaders think about AI readiness, much of the focus naturally falls on the information their organizations already have: consolidating systems, cleaning up data, digitizing archives, and making enterprise knowledge accessible to AI. But even if you get all of that right, new information will start arriving again tomorrow. Every email, form, document, image, and transaction has the potential to become part of the knowledge your AI relies on. Intelligent intake is about making sure that information enters the business in a way that makes it useful from the start, with the right validation, classification, and context attached to it. So, as AI takes on more consequential work inside the business, leaders must ask more than which models, tools, and agents to invest in. They must also ask: How confident are we that the information entering the organization is accurate, contextualized, and ready for AI to act on? Interested in Intelligent Intake? Bring us your most pressing questions and we’ll book some time to talk.

09.01.2026

We Asked a Room of Technology Leaders Where to Start With AI. They Wanted to Talk About People. 6 key takeaways that we heard during lunch The short version: Digital Intake, the point where information first enters your business, is the easiest and most important first step toward AI enablement. We hosted a lunch to discuss this. The room agreed, then gave us six better reasons for this than the one we walked in with. Earlier this month we invited a group of senior technology leaders to lunch in Calgary. They came from across industries, each of them responsible for the information their organization depends on. The premise for the event? AI is only as reliable as the information it can access, and many organizations have not done the foundational work of making their information trustworthy. “Digital Intake”, or standardizing what happens the moment information arrives in the organization, is where that work begins. Nobody argued with the premise, but twenty minutes into the discussion, when WCD’s Director of Digital Transformation and Innovation, Matt Christensen, opened the floor with what we thought was a technology question asking, “What does the first mile look like in your business,” the answer was not about documents. The first mile challenge is really around the culture shift and the change management. We talk a lot about AI, but I truly don't think leaders understand what that really means." And the conversation never really went back. Over the next hour, a table of people who manage information for a living kept returning to the same place. The hard part is not the technology. What they described, without ever using our language for it, was a very good case for starting at intake. 1. A process that works is not the same as a process that sticks The best story of the afternoon came early. During COVID, when everyone was sent home, one organization ran a beta of a digital mailroom. Everything was scanned. It worked. Until people came back to the office. What happened is when we started coming back to the office, the process broke because people got involved. And so, we had to resort back to the way we used to do things. It wasn't a culture shift happening." The technology did not fail. The business case did not fail. The process reverted the moment the external pressure that created it disappeared, because it had been adopted under duress rather than designed to last. A workaround survives exactly as long as the crisis that created it. An intake process has ownership, rules, and visibility, which is what lets it hold when the pressure comes off. KEY TAKEAWAY: If a process only exists because circumstances forced it, expect to lose it. Design digital intake as a standing process, not a workaround. 2. Build momentum by starting where people aren’t threatened If culture is the constraint, the practical question becomes where you can begin without triggering it. One attendee answered that better than we do in our own materials, then came back to it, unprompted, forty minutes later. The intake centre or the mailroom is like a gateway drug for AI. I can see it being a non-threatening way for business leaders to introduce AI processes. It's low risk. It's a great way to get the ball rolling. You get the organization consuming it, normalizing it, beginning to use it and leveraging it. Where it goes from there, those are internal decisions." This is the argument we came to make, put more plainly than we usually put it. Nobody feels endangered by better mail routing. The process work is close to identical to what you would do in a higher-stakes department, the risk is a fraction of it, and you finish with a working example and a real business case instead of a slide deck. Our CEO, Karen Brookman, made the same point from the other direction. Traditional back-office functions like mailrooms, may seem mundane, but they offer a practical, low-risk opportunity to demonstrate the value of AI-enabled digital intake. KEY TAKEAWAY: Digital intake is the lowest-threat place in the business to do real process work. The same discipline as a high-stakes department, a fraction of the risk, and a proof point you can carry anywhere else. 3. Everyone has an AI initiative. Far fewer have AI in production. David VanDerEems, WCD’s Senior Consultant on enterprise transformation, described a CIO conference where the room was asked who had an AI initiative planned for 2026. Effectively every hand went up. Asked who had an AI process actually running in production, he counted perhaps ten or fifteen. That gap was recognizable to everyone at the table. It is the same energy Karen hears from customers who tell her they have a hundred AI initiatives underway and no idea which one to finish. What closes the gap is almost never the model. It is the state of the information underneath it, and that state is decided at the point of capture. KEY TAKEAWAY: Digital Intake isn’t a later step in your AI journey, it is part of the foundation that determines whether AI can succeed. 4. The resistance does not start with the business The most uncomfortable story of the day is worth sitting with. One attendee had a ~500-page training document, scanned as a PDF, and wanted to make it useful. They loaded it into their AI platform and asked it to act as a document analyst they could put questions to. It hallucinated. Repeatedly. One of the pushbacks that I got from one of my colleagues is that it's hallucinating, you're wasting my time, you're wasting my efforts. Stop using AI." That pushback did not come from a skeptical business unit. It came from inside their own technical team. So the question they put to the room was if IT resists, what happens when we take it to the broader business? The answer from across the table was blunt and, we think, correct. People view IT or IT teams as really advanced and technology-embracing people, but they're not. People are people, and they like to stay in their worlds and do what they do." There is a second lesson buried in that story, and it is the reason we host these conversations. The model did not fail because it was a bad model. It failed because it was handed 500 pages of unstructured scans and asked to be precise. Nobody had ever decided what that document was, what was in it, or how it should be described. That is not an AI problem. That is an intake problem. KEY TAKEAWAY: When AI hallucinates on your own documents, the problem almost always arrived before the model did. Fix what happens as the information arrives in your organization and most of the trust problem goes with it. 5. It's difficult to explain an AI decision to a regulator The sharpest exchange of the afternoon lasted about fifteen seconds. Someone raised the obvious problem for regulated industries. Regulators audit your decisions. They ask you to show every step of your reasoning. And increasingly, a meaningful part of that reasoning is "the AI surfaced this." "So, justify that. Explain that." From across the table, without hesitation: "And you can't." Nobody in the room had a better answer. The consensus was only that this is coming, that it may end up being settled in courts, and that most organizations are not ready for the question. This is what digital intake is built for, and it is the part most easily overlooked. An auditable process, and a documented chain from where a document arrived, to how it was classified, to what was done with it. You are still moving toward AI. You can just show your work. Karen made the broader observation, and it may have been the smartest thing said all afternoon. Records management already solved this. Information governance has proven processes, and those processes gave people the confidence to say no to bad ideas. AI disrupted all of it and replaced it with experimentation. The answer is not to slow the experimentation down; it is to rebuild the governance underneath it. KEY TAKEAWAY: Defensibility is coming, and it gets built at intake. A documented chain from arrival, to classification, to action is what lets you show your work later. 6. Use the enthusiasm for AI to get the unglamorous work done Every other guest treated culture as the obstacle. One had inverted it, and this was the most practical thing we heard all day. Their organization asked each department to identify a use case where an AI agent could take over a taxing manual process. Departments got excited. Then they discovered that the information feeding those workflows was not standardized, and the agents could not work. It was this beautiful opportunity to go back and reinforce the necessity for foundations and create some standardization and structure. Because they wanted so badly for these workflows to work, we had their buy-in. We have a captive audience here." The foundations they went back to build are intake foundations. Consistent capture, agreed metadata, a standard shape for information arriving from a dozen different directions. Nobody gets excited about that work on its own merits. Everybody wants an AI agent. Key Takeaway: Use the appetite for the agent to build the intake foundations the agent depends on. Let them chase the agent and let the foundations get built on the way there. What are we taking from the day? Besides the honour and privilege of sitting in a room with incredibly smart and collaborative leaders, we went in to talk about information quality and came out with a sharper version of our own argument. The reason to start with digital intake is not only that it produces cleaner data for AI, though that remains true. It is that digital intake is the one place in the organization where you can do serious process work, build an auditable foundation, and prove the value, without anyone feeling their job is on the table. Culture is the constraint. Digital intake is the place it bends.

04.20.2026

Turning paper-filled file rooms into instant digital access for better banking experiences When this Canadian bank embarked on their digital transformation journey, they faced a daunting challenge—their branch network, spanning 260+ locations, relied entirely on physical records management. The Challenge: 110 File Rooms and Millions of Vital Records "This bank’s branch records were 100% physical," explains Mark Ellis, Senior Manager, Technology Services at WCD. "When you sign up for a mortgage, there’s a pile of paperwork that comes along with it. At the time, all of those mortgage files were being put in a physical file and stored in one of their 110 file rooms." These mortgage files weren’t just ordinary documents—they were vital records for the bank, containing highly personalized information that needed to be maintained with the utmost security and compliance. Some files were modest in size, while others were massive. "We had one mortgage file that spanned eight banker boxes," Mark recalls. "These weren't just for private individuals but also for business loans. At this scale, things can get complicated very quickly." At the same time, the bank was investing heavily in becoming a digital-first organization. They had rolled out a new SAP system that formed the backbone of their entire financial operation and were implementing Salesforce and Box as their enterprise content management (ECM) system. Beyond the Bottom Line: The Digital Banking Vision While the immediate benefits of digitization included reduced real estate costs, the vision extended far beyond simple economics. "The bank had made a strategic decision to reduce their physical footprint," says Mark. "People just weren’t going into physical bank branches like they used to." They recognized that mortgage documentation required substantial storage resources, making it an ideal target for digital transformation. But to truly cut back the physical records, they needed to scan them in full compliance with regulatory requirements, ensuring they would stand up in a court of law if needed. Untangling the Complexity Behind the Scenes The bank’s requirements went far beyond simple document scanning. They needed: Capacity to scan approximately 6 million images of mortgage files across their entire networkFull compliance with regulatory requirementsDocument-level scanning (not file-level) to enhance usabilityIntegration with their new ECM systemsAbility to accurately capture and display white embossing on white documents, ensuring critical details remain visible and legible But the real complexity lay in their document taxonomy: 256 distinct document types, each with unique business rules. "No single person could remember all 256 types and their associated rules," Mark explains. "For example, some document types needed to be physically retained because they contained original signatures that might be needed as evidence in court." When WCD first attempted to manually index a single box of documents following this intricate taxonomy, it took eight hours. “The person working on it essentially needed a PhD in the bank’s taxonomy," jokes Mark. Saving Countless Hours: The AI-Assisted Solution Rather than using artificial intelligence to replace humans—an approach that had proven unreliable—WCD implemented a hybrid solution where AI assisted human operators. "We took the philosophy of using artificial intelligence to assist the operator," Mark says. "The person makes the decisions, while the AI system does all the basic work." WCD's Kofax enterprise imaging workflow system was programmed with all of the bankB's taxonomies and business rules. It would read through each document, classify it according to document type, and then guide the operator on what needed to happen with that document. "That's how we took something that initially took eight hours for one box down to minutes," Mark says. "When we demonstrated the system during user acceptance testing, their team was surprised by two things: one, that we got it right, and two, how efficiently we were doing it." Same-Day Digitization: How New Mortgage Files Are Handled The partnership evolved when the bank expressed interest in deploying similar technology within their own operation. WCD's solution was to provide their technology on an as-a-service basis—meaning the company never had to pay for licenses or capital expenses. "While we were scanning the bank’s mortgage collection, the same back-end infrastructure was also being used by their own people for other types of records," Mark explains. "This gave them a choice. They could process some documents internally and send others to WCD. It handed power back to their team instead of them being locked in by technology and forced by vendors. That's the essence of a true partnership." The arrangement created a natural division of labour. The bank handles daily processing of new mortgage files in-house, while WCD manages large backfile projects when branches are closed and their document repositories need to be digitized. Building the Bridge to Box with Custom Code During the project, the bank initially requested a six-month delay to allow their IT team to develop a solution for integrating the scanned documents into Box, their enterprise content management system. WCD saw an opportunity to do it faster, better, and more cost effectively. "I said if they could give us one month, we could deliver direct integration," Mark remembers. "Two weeks later, it was complete and we rolled it out." This earned significant credibility with the bank and demonstrated the value of WCD’s proprietary software backbone, developed over the previous decade to sit atop their Kofax system. "It allows us to write small amounts of code and plug and play," explains Mark. "We don't need to create the entire system for each client—we're just adding that final piece to connect with their environment and implement their specific business rules." The integration WCD developed is sophisticated and fully automated: Every night, the system automatically packages everything scanned that dayAutomated checks verify the content matches the initial transmittalThe system logs into the bank's Box environment through a secure linkIt searches for each client by identification numberIf the client exists, it loads each document in its corresponding areaIf the client doesn't exist, it creates a new customer folder from a template and then loads the documentsAll metadata and audit trails are properly maintained throughout The system accounts for approximately 2 million of the bank's annual 8 million API call allocation with Box—and even that figure is optimized through WCD's efficient algorithms. One Partner from File Retrieval to Recycling A key advantage of WCD's approach was taking full responsibility for the entire transformation—from files on the shelf to digital delivery. By contracting with WCD for the complete service, the bank streamlined the process, which includes: WCD records specialists retrieving documents from branches and offices with full chain of custody to uphold regulatory complianceScanning and processing according to the bank's complex requirementsDigital delivery directly to Box with complete metadataProper handling of physical documents afterward—with approximately 20% needing long-term physical retention and the rest undergoing secure destruction "The paper isn't destroyed, it's recycled confidentially," Mark explains. "It goes through a process that turns the paper back into pulp, washes all the ink off, and it becomes tissues, paper towels, and toilet paper. We've satisfied environmental concerns while ensuring the safety of people's personal information." The Real-World Impact: Better Service, Flexible Work The impact of this digital transformation extends beyond cost savings and efficiency. It has fundamentally changed how the bank serves its customers. "When someone calls them regarding their mortgage, the person on that call line can find the exact information they need within minutes," Mark says. "They can speak intelligently about the customer's situation without having to say, 'I'll call you back when we can get the information.'" This transformation was particularly valuable when COVID-19 hit. Because the bank had already digitized so much of their operation, they were well-positioned for remote work. "They were already far ahead of the curve," notes Mark. "The pandemic had less of an impact on their operations because their executive had the foresight to make these changes. Knowledge workers still needed access to information, and through projects like this, they had eliminated the paper dependency that would have made remote work much more challenging." Beyond Banks: Lessons Worth Learning This bank's journey with WCD demonstrates these key principles for successful digital transformation: 1. Think beyond scanning Document digitization is about more than creating electronic copies, but rather transforming how people work with information. 2. Embrace artificial intelligence The most successful implementations of AI enhance human capabilities rather than trying to replace them. 3. Choose partners, not just vendors Impactful technology implementations come from true partnerships where both parties are invested in finding the best solution. Through their partnership with WCD, this bank has not only modernized their document management but positioned themselves for the future of banking—one that's digital-first, customer-focused, and agile enough to adapt.