Transcript: 

 

Pete Wright: 

We swim in an ocean of data, yet we thirst for knowledge. It’s a paradox of the digital age. We’ve built libraries the size of continents filled with every conceivable bite of information, yet finding the single crucial fact, the precise piece of wisdom we need often feels like searching for a lost grain of sand. Why? Because information isn’t knowledge, it’s raw material. And just like raw iron ore needs to be forged into steel to become truly useful, information needs to be curated, refined, and delivered precisely when and where it’s needed. 

Welcome to Connected Knowledge, I’m Pete Wright. Today we’re diving into the deep well of knowledge curation and delivery, exploring how AI is transforming the way we learn, work and thrive in this age of information overload. Our own Fiona Vanderlinde is back with me today to help us unlock the secrets to turning data into actionable knowledge. 

Fiona, welcome back. 

Fiona Vanderlinde: 

Thanks, Pete. It’s lovely to be back. 

Pete Wright: 

Well, it’s lovely to have you and we’ll start with the drowning question. How about that? We are drowning, we’re drowning in data and starving for intelligence. How are organizations navigating this particular paradox and actually figuring out how to use information that they’re accumulating? 

Fiona Vanderlinde: 

It’s really, it’s an interesting fact because just looking at it, at the moment, I’d say over 90% of the world’s data has been created in the last two years. So with the advent of AI, everybody’s creating endless amounts of data. But we’re not just talking about all of that external information, it’s internal knowledge bases, which are also growing exponentially as well. So to your point, it is information overload, everybody is drowning. And even though we’ve got all of this extra information at our fingertips, it’s, do you have the right information at the right time for the right audience and the right channel? So- 

Pete Wright: 

I just want to make sure I’m not crazy. You said 90% in two years? 90% of the world’s information has been created in the last two years? You said that out loud, right? 

Fiona Vanderlinde: 

Yes. I did, yes, and so we’ve now got AI training off AI created content, because it has run out of content to learn from that humans have created. So we are in the exponential information overload stage and there is so much information out there that if you just think of your customer center agents, they’re drowning. There is emails, there are chats, there are people next to you if you happen to be working in an office still. How do you know what’s right? 

Pete Wright: 

Well, I mean that’s the existential question of the AI age, right? How do you know? I’ve got a lot of questions for you, but I feel like you poked the dragon with this particular question. 

Fiona Vanderlinde: 

I certainly do. 

Pete Wright: 

How do you know… right. We’re hearing more and more about AI slop and I think you just described it, AI training on AI, right? 

Fiona Vanderlinde: 

Mm-hmm. 

Pete Wright: 

This cascading sort of dependency chain of AI creating more and more information that is of potentially a wide array of trust factors against it. Right? I am super curious how we could possibly imagine that we have a solution to help these agents get over or through the bog of slop. Doesn’t the deluge of information impact just agent-level productivity and decision-making? 

Fiona Vanderlinde: 

It certainly does, and if you think about yourself as an agent and you’re trying to understand and answer a question for a customer, but even today within agents, it’s not a like-for-like. So back in the day when I was an agent, it was all voice interaction. So I knew that I was only going to get interrupted via voice and that was the only channel I had to answer. And so all of my answers were in my knowledge base and I was able to go to this one large repository. I mean, I don’t want to show my age, but mine started as a book on my desk. So I mean, thankfully- 

Pete Wright: 

I’m going to drop the word sheet protectors. Does that make all the hair stand up on the back of your neck? 

Fiona Vanderlinde: 

Oh, it does, indeed. So yeah, knowledge has existed in one way or another throughout the years and thankfully now we’re mostly all online, virtual, although there’s still a lot of people who may have some paper-based stuff out there. But I guess once you move to virtual, the number of documents exponentially grow and grow, and on top of that, the agent’s life and how they need to multitask is also growing. So I mean, like I said, it used to be voice where you have to answer that call within so many seconds and respond appropriately, but now we’ve got live chat where they have to also handle other types of conversations and often more than one at a time. 

Then you’ve got email queries where you need to give really detailed correct consistent answers, social media interactions, which just pose their own whole type of issues I guess, and you need a whole different tone and approach for those. We’ve got video chat. I haven’t seen that as much, but this definitely adds another whole visual dimension to knowledge delivery. And SMS, like go back older school again. And so each one of those channels, they have their own unique requirements and what works for one won’t work for the other. So you need to really work it out. So you add all of that to the mix of, plus I’ve got all of this information everywhere. How on earth are people going to work out what fits where? And I guess what we’re trying to say is, they shouldn’t have to, it should be delivered to you in a way that makes sense. In a way that kind of removes that destruction and that overload and the burden from our agents having to manage all of that.   

Pete Wright: 

Right, if I imagine our agents are surgeons, right? Your agent is a surgeon and let’s just say they need to know, do I cut that thing right there or this other thing over there? Note my ignorance by just calling anatomy, things. A generic manual doesn’t really cut it. So here we are, we’ve got to transition into the contextual relevance part of our conversation. How do you figure out how to take the morass of information and rediscover contextual relevance across all of these channels? And of course, I’m not asking you for the recipe, but just a, your sort of ideology around this thinking from an agent forward perspective. 

Fiona Vanderlinde: 

Yeah, definitely, and I think when you’re talking about the recipe, you’re very spot on there. So we really think about it as three key dimensions. So it’s the channel, it’s the audience, and it’s the situation, and I feel like it’s the situation that often gets missed. 

So when you think about your audience, mostly we’ll be talking about an internal audience, your contact center agents, and they need that deep domain detailed knowledge, which is very clear, process driven, has troubleshooting steps, things like that, maybe scripting. And then also if you think about their experience, they may prefer to be able to quickly access that information via a chat, opposed to reading through and following the bouncing ball. So how do you give them that type of information? But then you’ve created all of this information or knowledge and you might want to also share that to your external audience, like your customers through a self-service channel. And one of the things we haven’t talked about yet is the introduction of AI agents. And so they’re a real thing and they need knowledge in a very different way again. They need it structured, machine-readable formatted. And so the beauty of a modern knowledge management system is that you can take all of that core knowledge assets that you’ve created, spent all your time on. 

Let’s say for example, I’m going to switch to a banking scenario or a purchase scenario, rather than medical, and just say, let’s say we’re going to process a refund. And how do we automatically transform that information for different users and use cases? So for your agents, they’re going to see that full procedure with all of the internal systems they need to access and any compliance processes. And then maybe for your experienced agents, they’re going to access the information via chat and it’s going to be presented to them as a direct answer to the actual question, rather than the whole document. And then your customers are going to see a simplified self-service version, your AI chatbot gets a structured machine-readable format, and your mobile app gets a streamlined step-by-step guide. So it’s all about making your knowledge work harder for you and turning what’s generally considered a cost, into an asset. And I think that’s really key. So create once, deliver everywhere, but in a way that makes sense for each of the contexts that you’re delivering.  

Pete Wright: 

Yeah, I mean, you just got to where my head was going. And my next question is all around one size fits all doesn’t generally fit anyone. And so when you’re looking at AI to help sort of tailor knowledge delivery to individual channels and across learning styles, learning styles and needs, here we’re talking about creating a truly personalized experience. How far do you think AI can take us down that road? Are we talking about multilingual at this point, that we trust a multilingual delivery? How far can we go? 

Fiona Vanderlinde: 

So you can do anything at the moment. I think your question of trust is the biggest one there. So yes, you could most definitely get AI to do a transcription of all of your knowledge and then present that out. But as we know with any form of transcription, unless you’ve got a human over the top, it doesn’t really hit the ability to get the regional dialect or common terms and phrases. Like even in English, I mean, you’re in America, I’m in Australia, I talk to Americans all day and some of the words and things that I say, they take a step back and say, “Oh, I’m not sure what that is.” So we’re all speaking English and yet at the same time, we’re not at the same level of English as each other. So when you’re asking AI to transcribe something, it’s going to do it in the most formal, predictable way possible and it’s not going to hit the mark all of the time. 

So yes, you can, but I always say just because you can doesn’t mean you should. So let’s try to rethink about what makes sense. So yes, definitely get it to do that. It can re-transcribe all of your information into a different language, but get a human over the top just to make sure that it hasn’t misrepresented something that you’ve written in the wrong way. So let’s definitely use it to save time, but don’t rely on it to do end to end. 

I also wanted to touch on when you talked about the personalization. So that’s a really key part of knowledge delivery. It’s a lot more than just adapting content for your different channels and audiences. It’s about thinking about the personas who are going to be accessing that knowledge. So I mentioned before around agents, imagine you’ve got a brand new agent, they’ve literally just walked into your office and they need all of that additional context and explanation. In the knowledge industry, we call that assumed knowledge. So a lot of new agents who are reading things, they have all this assumed knowledge that the experienced agents just assume that everybody knows. And it might be as simple as after you’ve typed in this thing, you have to press enter or you have to click that weird little button that doesn’t really make sense. Agents who are new need all of that explanation, all of that additional context, versus the experience ones who I said just need a quick reference point or a fee or a bit of information to answer a particular question. 

And then we’ve got, sorry, technical specialists who need deep, detailed information. Team leaders, they need both to know how to support their agents, but also how to coach them. We need the customer who needs that clear jargon-free guidance, and the AI system who needs, like I said, the structured machine-readable content. But each of those users, they come with their own background, experience level, and like you said, learning styles. So a truly intelligent knowledge system recognizes these differences and they can adapt or present the knowledge accordingly. 

So like I said, new agents might see all the additional explanationary notes and system navigation help or scripting how to answer a phone, because when you just put on those phones for the first time, you’ve gone through all of this training, but you might answer the call and forget what your own name is, let alone what company you are suddenly working for. So let’s give them all of the help they need to provide that great customer experience. 

Pete Wright: 

It’s funny that you bring that up, especially the user interface parts, which can be deceptively tricky. I was speaking with somebody, an HR call center manager who said, “Our biggest struggle in hiring new people is that we’re hiring younger and younger people.” And the biggest challenge we have is our system is chock-full of disclosure triangles. Little triangles that you have to click to reveal more information, and all of them use their phones. They don’t use PCs, they don’t understand the big screen at the level of intimacy that some of our older agents do. And they have to be trained on user interface specification that we, as you say, assume is standard that we all understand how this works. It doesn’t always work that way on a phone. Frankly, maybe we’d be more efficient if we just gave all of our new call center managers phones and said, on mobiles and go, “Go for it. Go do your worst. Let’s see what you can do.” 

Fiona Vanderlinde: 

Most definitely. Yeah, and we find the same. Even I’ve talked to people when I was showing them an icon and it was a little envelope, which I don’t know, means mail for me because it’s an envelope. 

Pete Wright: 

And me. 

Fiona Vanderlinde: 

And the newer people, they’re like, I don’t even know what that is, because they don’t use posted letters anymore. And so the iconology that’s used, how do you make that work across multi-generations and multi-people? It’s a real problem these days. 

And I know what you mean about the mobile friendly, like everybody wants it in that small, sharp little boxes, they don’t want all the long, boring documentation, all of that information. So it is a real challenge. And I think more and more it’s around, how do you write for this newer generation to keep them engaged? I don’t know how true it is, but I’ve heard that the attention span of the newer generation has decreased even more than what it had previously been. So you’ve got literally seconds to get their attention and get them to read and follow this information. 

Pete Wright: 

Side quest, we’re going on a side quest, you and me for a minute. 

Fiona Vanderlinde: 

Okay, cool. 

Pete Wright: 

Have you played with Google’s NotebookLM yet? 

Fiona Vanderlinde: 

No, I have not. Why? 

Pete Wright: 

Okay. Look, I know, you work in a space where your mind is probably impossible to blow anymore, but let’s try and set you up to have your mind blown. NotebookLM, it’s a consumer product. You can go into it and sign up for the beta now, and you upload your resources to it. PDFs, websites, documents, a notebook app, and then you can have it generate a podcast where two people are talking about your content, interviewing one another about your content. And I am a podcaster, Fiona, and I was tricked. 

Fiona Vanderlinde: 

Wow. 

Pete Wright: 

I was blown away at the kind… Now, is it perfect? No. Was it uncanny valley, just legitimate? Did it talk about my content? Yes, it did. Can you make it hallucinate? Sure. But is this the future of content delivery across modalities and learning styles? Oh my gosh, yes. 

Fiona Vanderlinde: 

Yes. 

Pete Wright: 

Oh my gosh. 

Fiona Vanderlinde: 

Definitely, and I mean that, okay, that has blown my mind. I’ve heard of people being tricked with getting a phone call and you’re having a whole conversation, it’s not towards the end that you realize that you’ve been talking to a bot the whole time. I’ve heard of that and that’s really clever and blows my mind as well, and but wow. Okay, and I think you’re right, the multimodal, that is a whole thing of the future as well. So we always talk about knowledge being delivered via text. Does it always to be like that? And maybe for contact center agents, potentially, but if you are somewhere where you could get it talking to you in your headset while you’re fixing an engine, for example, or hopefully not while they’re operating on me, but operating and reconnecting that thing to the other thing. 

Pete Wright: 

Somebody else. 

Fiona Vanderlinde: 

Yeah, somebody else, not me. I don’t want to trial that one, but you’re right. 

 Pete Wright: 

Yeah. Yeah, and the fact that it generates it in 60 seconds, like this new way to approach our content is fascinating, so. 

Fiona Vanderlinde: 

It really is, and I guess just if we think about the role of AI within knowledge management, it definitely is transforming the delivery. And so what we’re finding in Upland Panviva is that the content transformation around automatically rewriting agent-facing knowledge into customer-friendly language. As a knowledge manager, I’ve worked in this field for way too many years and I was trained how to write knowledge for agents or internal staff. I haven’t been trained how to write it for customers. And so when presented with that, that was really challenging. Having AI come in and set the tone and redo it for me straight away and I just have to do a couple of tweaks, life-changing. Absolutely made it so much easier. You look at those long form boring, I’m saying boring, but I know that it’s really key information.  

Pete Wright: 

Right, right. I mean, it’s okay, no take-backs, but it’s okay. 

 Fiona Vanderlinde: 

It is, but I mean, you open a document and it’s 60 pages long and you just think, oh, how am I going to read that? I don’t have the energy for that. But getting AI to, like you said, go through all of your information, condense it, and break it into chat friendly snippets. You can even get it to do text to voice. So rewrite it from the text that you’ve got and put it into a voice, that’s very possible as well. And just creating that structured data that your AI chatbots need from the narrative content that you’ve already got today. 

So, it can also do really cool things like add all your metadata and classifications, the auto tagging of the information so it’s easier to find and consume, and the contextual intelligence. So reading all of that interaction content and it can even predict knowledge needs. So we’re seeing a lot in the CCaaS, or the contact center as a service space, you’ll see a lot of that where it’s grabbing all of that contextual information from the customer and the agent and then doing a predictive pop of what it thinks you are going to need next. So yeah, it’s mind-blowing. 

 Pete Wright: 

Okay, mind-blowing, all right. We’ll call the side quest concluded and now we will transition into some new stuff that we want to talk about. Let’s talk about smart snippets. Paint a picture for our dear listeners. What are they? How do they cut through the noise to deliver what we need and right when it’s needed? Smart snippets. 

Fiona Vanderlinde: 

Yeah, smart snippets. So it’s a really cool term that Panviva is delivering, and it’s around really revolutionizing how we think about knowledge delivery. So instead of forcing users to wade through that entire document, smart snippets, it delivers a bite-sized, contextually relevant piece of information exactly when that agent needs it or that consumer needs it, really. 

So think of them, and this is going to go really old-school for Panviva is, you think about them as your organization’s knowledge GPS. So they don’t just show you that entire map that then you have to try and work out where the heck you are. They guide you turn-by-turn, exactly where you need to go. And we’re seeing organizations who use smart snippets be able to deliver faster time to competency for new employees, a reduction in errors, and just improve customer satisfaction because they don’t have to keep asking the same questions over and over and over again. And for the agents, it really means that no more tab switching or document scanning during customer interactions, or worse yet, dead air or putting those customers on hold. And for the customers it means they’re getting that accurate consistent answer across all channels. 

We’ve got some customers who are using the smart snippets today and they’ve gone from having just one voice channel to delivering content across five different channels in their organizations. So they’re really getting the most out of their knowledge. They’re not just using it for the one channel, one audience anymore. 

Pete Wright: 

Well, and this gets to one of the things we’re talking about front, which is this idea of we’re not just talking about sort of AI writ large, we’re talking about using these tools in a clever way to repurpose our content in a way that does address our core metrics. 

Fiona Vanderlinde: 

Exactly, yeah. So it’s using, so if we get into Upland Panviva, we’ve got a system called Digital Orchestrator. And Digital Orchestrator is what creates and pushes out the smart snippets to all the various channels and audiences. And so if you think about it, orchestrator, along the whole musical theme, think of it as the conductor of your knowledge. And so it’s really ensuring that each one of your instruments, or in this case we’re thinking of pieces of knowledge or snippets of knowledge, plays in perfect harmony. 

So it uses AI to automatically categorize and tag the content, add all the keywords, also grab the content based on a particular question or answer question response that you’re trying to provide back, and be able to rewrite it through all of the content that you have to deliver the answer to that question. So it really ensures the consistency across the channels and it delivers that personalized knowledge experience that we talked about earlier. And a lot of that is with AI, you can definitely still do it all with just no AI and just use a human, but what we find is that by implementing the AI features, it means that you’ve still got the human in the loop because they’re the ones going in and reviewing, approving it, and then doing the ultimate publish. So it means that when you are pushing that content out to your customers, you know that it’s right. You’re not going to have as many of those hallucinations or AI errors that we’re unfortunately seeing more and more in the media these days with companies being bitten. 

So all of the knowledge that they’ve created for their internal agents, they’re able to deliver to their customers and via all of the different channels online. So what it’s meant for them is they’ve been able to reduce, I guess, some of their FTE for their agents by implementing self-service for their customers. So we know that in the world today as a customer, I don’t want to have to always ring up a contact center to talk to somebody. I want to be able to at any time of day, go online and get an answer to what I think is a very basic question. And so what you’re doing by using Digital Orchestrator and providing smart snippets is giving that aware and compliant information to those customers at their time of need. 

Pete Wright: 

It’s incredible technology and I have to say, the more we talk about it on this show, this is going to sound sort of sadistic, the more excited I am about having problems that I can call a customer service agent and just see how they’re using it. Is that, that’s- 

Fiona Vanderlinde: 

No. 

Pete Wright: 

That makes me sound like a creep. I want to start breaking my own things just so I can have an excuse to call a support agent. 

Fiona Vanderlinde: 

Well, it’s difficult because I feel exactly the same, like whenever I call up your bank or electricity provider, I expect this level of service and when I don’t get it’s really frustrating. And I know it’s not the person on the end of the phone’s fault, but it’s like, why isn’t your management helping you to help me? It just doesn’t make any sense, so. 

Pete Wright: 

Right, okay. Well, look, we’re going to talk about what folks should do next in a minute, but we have listener questions. And it’s been a while since we’ve taken on any listener questions. I am very excited that we have two that are right up your alley today. And so we’re going to throw them at you and see what you think. 

Number one, this one comes from someone who would like to be left anonymous. With all the buzz about AI, how do we find that sweet spot between using bots and keeping those empathetic human agents for the tricky stuff? I have leadership that wants to go all-in on AI and we’re struggling to find a balance that addresses those of us with real concerns. 

Fiona Vanderlinde: 

I think this is perhaps the most pressing question in customer experience today, and I’m really glad that the listener is focusing on the balance, rather than the extremes. So we hear this question all of the time, and it’s from both ends of the spectrum, really. So let me talk about something that hopefully will help. 

So if we think of AI and human agents as different instruments in that orchestra, rather than competitors for the same role. So AI really excels at handling those high volume, repetitive tasks with a very consistent answer. So things like, I want to check my order status, or I want to process a simple return, or I want you to give me some basic product information. So this actually really elevates the role of the human agents, rather than replacing them. So what we’ve seen is successful implementations of a complexity-based routing approach. So it’s where you get AI to handle all those transactional, simple queries, and that then frees up your human agents to focus on what they do best, which is handling the complex issues that require the empathy, the judgment, and that creative problem solving that I think humans can do really well. 

And for example, when a customer is dealing with a combination of issues or they’re experiencing emotional distress or when the situation requires an exception handling, so outside of the norm, these are the scenarios where human empathy and experience are irreplaceable and AI just simply cannot come in and give you a great experience there. 

So the key really is, how do you use your knowledge management system to support both channels effectively? And your core knowledge really needs to be structured so it can serve both the AI and the human agents and ensure that it’s a consistent approach, while adding a human touch to what matters most. So there’s no point in creating an AI journey that your customers do not want to use. So it’s really about thinking upfront and doing a lot of planning and conversations with customers, looking at your complaints, looking at your call logs to understand, what can we automate and what do we need to keep as human? 

And I guess for those who are concerned about going all-in like our listener, I would really talk to your leadership about the metric around, that organizations, they need to maintain a balanced approach. So that’s really where Upland comes from. It’s about using AI for a routine task while elevating human agents to handle those complex interactions. And by doing that, what you’ll typically see is higher customer satisfaction scores and a better resolution rate than those who go in either direction too far. And so really, it’s the sweet spot, it’s not about choosing between AI and humans. I think you said we’re pretty much already at the stage where AI is here to stay. It’s about thinking about which one is best to use where, so that you can get them to work together, rather than competing or just being one or the other. So, simply what I’d say. 

Pete Wright: 

Simply? That was terrific. You’re very good at this. I’m going to give you another question. This one is a little bit different, it’s a policy question. And let’s just say I see why maybe they wanted to also remain anonymous. How do you handle reps who potentially inadvertently leak company proprietary information in a call situation? We are trying to build a new policy around two separate incidents and would love some assistance. Where do you start? 

Fiona Vanderlinde: 

Wow. Yeah, that’s a really, really tricky one, but it’s a great question. And I think there’s a lot of compliant organizations who really stuck with that. So what I would say is that the key is to create a system that really prevents the leaks or that getting the information out there, rather than just reacting to it. 

So what I’d recommend is to use your knowledge management system, and that should be your first line of defense and implement some real clear visual indicators for what is internal only versus what is customer facing. So make it simple for your agents to know the difference. You could also, if we’re talking about smart snippets, you could implement those, and so that way it’ll automatically only provide the customer appropriate version of internal information. And when you’re thinking about how you’re delivering your knowledge to your agents, if they’re working in their CRM like Salesforce or ServiceNow or things like that, make sure that if they’re staying in there and they’re looking at the customer information that again, they only see the approved customer facing information. So it’s really reducing the risk of compliance failure, rather than just hoping for the best. 

But most importantly, like I think our listener’s already doing, is use the incident as learning opportunities. So what I’ve usually seen is that it’s rarely a case of deliberate policy violation or the agent trying to do the wrong thing. It’s usually a system gap or a process gap, and those things are great because we can actually fix them. So the key is really to make it a lot easier for the agent to do the right thing rather than the wrong thing, and your knowledge management should be an active participant in preventing these incidents. So don’t think of it as just a passive repository of information, it is definitely your key to fixing this. 

Pete Wright: 

Out of the park, Fiona, out of the park. 

 Fiona Vanderlinde: 

Thank you. 

Pete Wright: 

And you brought it back to smart Snippets, that’s my favorite thing of the day. Look at that. 

Fiona Vanderlinde: 

I know. 

Pete Wright: 

Hey, where would you like people to go learn more about Panviva, Digital Orchestrator, Digital Conductor? That’s just going to be what I call it now, and Smart Snippets and all the cool things you can do about it? What’s the next step? 

Fiona Vanderlinde: 

Well, what I’d really like to see is if you’re interested in anything we’ve talked about today, go to the Upland Panviva website and you can book a demo and we’ll be able to show you exactly how Digital Orchestrator can transform your organization’s knowledge management strategy. And just remember, it is a digital age, so it’s not just about having knowledge, it’s about delivering the right knowledge to the right person at the right time in the right format. And that’s exactly what we do with Panviva’s Digital Orchestrator. 

Pete Wright: 

Outstanding, we’ll put the link to that in the show notes. 

Fiona Vanderlinde: 

Thank you. 

 Pete Wright: 

Thank you so much, Fiona, for your time today. 

 Fiona Vanderlinde: 

No problem, it’s a pleasure, Pete. Always lovely being here with you. 

 Pete Wright: 

Back at you. Anytime you want to come back, the floor is yours. Fiona Vanderlinde, Upland’s own. We are thrilled to be able to share this conversation with you. We appreciate your downloading and listening to the show. Thank you for your time and your attention, and we’d love to hear from you. Just again, swipe up on the show notes and look for that feedback link to send questions to us or any of our past guests, and we’ll do our best to get them answered just like we did today. 

On behalf of Fiona Vanderlinde, I’m Pete Wright, and we’ll see you right back here next time on Connected Knowledge.