Episode Transcript
[00:00:22] Speaker A: Hello, everybody, and thank you for tuning in to this episode of Metrogy's Metrosite.
I'm Beth Schultz. I'm the vice president of research and principal analyst at Metrogy. And with me today is Whitney Mirror. Whitney is an applied AI product manager with Nice, and we first met at the future of CX track that Metroji hosted at the IT Expo supershow in Florida earlier this year. Whitney participated in a panel on knowledge management, and she had some really great insights to share. So I wanted to invite her here for the Metrosite podcast as well. I mean, certainly we need to talk a lot about knowledge management these days, given how vitally important it is to AI, and particularly regarding AI enabled customer experience. So I'm really happy that Whitney took me up on my offer. Whitney, nice to connect with you again, and welcome.
[00:01:18] Speaker B: Thanks, Beth. It's great to see you and great to be here.
[00:01:22] Speaker A: Good. Okay, so let's just start with some fun facts with you. A little bit personal, a little bit professional. Just tell us three fun facts about yourself outside of the work world.
[00:01:34] Speaker B: Yeah. I'm a textile artist, and I teach embroidery at a nonprofit in Denver called Recreative.
I'm a cocktail snob, and I actually have tattoos of cocktail olives because I just love them so much. And for about three months last year, I ran an e commerce shop that sold soup themed hats.
But I shut it down.
I did not embroider them myself. I thought about it. Yeah, it was a drop shipping situation, but my shopify free trial ended and I was not turning a profit, so I shut it down.
[00:02:16] Speaker A: And what is your favorite cocktail?
[00:02:19] Speaker B: A dirty martini.
[00:02:20] Speaker A: A dirty martini. Oh, I love a dirty martini myself. So that's why you have the olive tattoo.
[00:02:26] Speaker B: Yeah.
[00:02:26] Speaker A: Okay. And textiles. What kind of textiles do you embroider? What can you tell us a little bit more about?
[00:02:31] Speaker B: Yeah, actually, you can see one in the background back here. Just kind of wall art and pillows, napkins, stuff like that. It's very soothing and meditative.
[00:02:44] Speaker A: Nice, nice. Great. Okay, so let's talk a little bit more on the professional side. Your title is applied AI product manager. What does applied AI mean? Broadly, and then specifically in your role at nice?
[00:02:59] Speaker B: Yeah, it's kind of a new title that I think I've seen people sort of take up. Ultimately, it's a product management role. My role is to build, help ensure that the team builds delightful software that helps our business grow. And then applied AI is the descriptor of the technology that I hope to use to meet that brief AI.
All kinds of technology, all kinds of folks building models, building the technology of AI and applied AI looks like integrating that technology into existing tools and workflows. So that's sort of broadly what it means.
And then specifically at nice, I'm on the knowledge management team. So I'm bringing all of the wonderful generative AI technology we're seeing, all of the analytics capabilities in AI that we're seeing, and trying to apply it for our knowledge management product.
[00:03:55] Speaker A: Excellent. Okay, so your product manager, did you have an interest in AI prior to getting the product management role?
When did you start seeing it as a career opportunity?
[00:04:08] Speaker B: Yeah, it's a great question. I started in the world of data pretty early on in my career. I was a customer success manager at a tech startup, and I could see how much data we had at our fingertips about our customers that I thought we could use to make our customers lives better and our business better.
And then when we hired a data science team at that tech startup, it just blew my mind, the kinds of things they were able to do with that data. So I had the intuitive sense because I could see it. And then I saw what the really skilled practitioners were able to do, and I ended up leaving that job to do a data science bootcamp. And I learned to code and I learned how to build models. This was years ago at this point, so my coding skills are a little bit rusty, but I really learned the fundamentals of how this stuff works, how the technology works, and it just was, it sort of all organically came together to be the basis of my career. And then I've worked at AI startups, and then I now obviously work at nice, doing this applied AI stuff.
[00:05:13] Speaker A: So it's interesting because you have such a data centric, heavy, geeky kind of professional role, and yet you also have this very artistic side, obviously.
[00:05:23] Speaker B: Yeah. Gotta have balance.
[00:05:25] Speaker A: Absolutely. Okay, so you explained a little bit that you're working on applied AI rather in context of knowledge management products. So can you tell us a little bit more about how knowledge management fits into your role?
[00:05:40] Speaker B: Yeah, it's the foundation of my whole role. So the product I manage is called CX, one expert, which is one of the knowledge or the knowledge base product at nice.
And so everything we do is focused on building delightful software for our knowledge managers and trying to make sure that their lives are easy and efficient and that they get the knowledge that their customers need in those customer experience workflows.
[00:06:08] Speaker A: Okay, so knowledge management, obviously, that is just kind of a very old thorny problem. There's no surprise there right. It is one of those tricky things that companies are challenged with.
And because it can be so hard to do well, you know, knowledge management disciplines in many companies very, you know, far from rigorous, which has really sort of exacerbated, been exacerbated with the introduction of AI. So from a technology perspective, what does sort of strong knowledge management look like today for customer experience purposes versus weak knowledge management?
[00:06:52] Speaker B: Yeah, it's such a good, it's an exciting time to be in the knowledge space because AI is really, it really shines a light on your knowledge management process. If you have a strong knowledge management process, you're able to just scoop up an AI platform, start running with it, enable your customer experiences with AI, and if you have a weak knowledge management process, you are going to see just challenges adopting AI. So when we see strong knowledge management practices and teams that are really able to just start running with AI, what we see is they've got web native content that is structured effectively. So it's in some kind of information architecture, information hierarchy that makes sense. And it's published on the web, even if it's behind a login for your employees only, it's not PDF's in a sharepoint somewhere completely free of hierarchy.
It's not five wikis that people just keep adding to without editing and curating. So that's one piece of it. The second piece of it is folks have a culture and a practice of looking at the data, looking at usage data, seeing who's being served by their content and their knowledge and making it better, and a practice of continuously improving it based off of metrics and data. And the third thing that we see for a strong.
Oh no, there's not a third thing. Let me back up.
[00:08:33] Speaker A: So it's, well, it's structured content, the people in the processes around that technology, right? Yeah. Okay, so what if my knowledge man systems today, they're outdated? How do I get started down that right path?
How big of an undertaking is this?
[00:08:56] Speaker B: Yeah, I think people are often overwhelmed by it, and it can be emotionally daunting to look at it and say, how do I get there?
The good news is that if you work with a vendor who knows what they're doing, they'll be able to walk you through the steps to go from where you are to where you need to be.
And I'll just brag about the team at expert for a second. I think we have just a world class onboarding team who can really help you identify who your core users are, what your structure needs to be in order to support them. And some of what we call our cornerstone content that will serve the needs of those users, the needs of that audience sort of, to start out, and they'll really help you build that strong foundation. They're just, they're so good at it. But if you talk to them, yeah, they'll tell you there's three things you need. The people that you're serving, your audience, the structure to build out your content, and then your cornerstone content. So the things that are going to meet the needs of the people who you're trying to serve. And as much as you can distill that audience into a very clear cohort of users or of customers, the better off you're going to be. It doesn't have to be small in terms of numbers, but it does need to be distilled in terms of their needs and use cases to start. And then you build that muscle with the small group and you can expand out.
[00:10:25] Speaker A: So you start with a small group, one group, or you make sure that you have distinct groups.
[00:10:34] Speaker B: What we see in our customers is a phased rollout. Works really well. So you start with your one group and you build up your muscle to serve that group. So maybe that's all your agents, maybe that's all your customers of one product, and then you can roll out to more and more and more.
Or maybe that's all your product documentation or all your support documentation, but it's really about defining a cohort that you can serve really well to begin with.
[00:11:08] Speaker A: Okay, so let's talk a little bit more about knowledge management and sort of that rapidly stating, sorry, rapidly evolving state of AI. What really do companies need to understand about what's happening today?
[00:11:24] Speaker B: Yeah, I think folks need to understand that it's very noisy out there in the world of AI, but solid knowledge management practices still apply. And what we've seen over the last couple years, that has been true as this AI boom has occurred, and I think will continue to be true. If you have readable, structured, discoverable content, your customers are going to be happy with it, your agents are going to be happy with it, and then the AI is just going to be able to slide right in and have success with that content as well.
[00:12:00] Speaker A: Okay. So there's nothing's really changed. It's just now the imperative to do it, right.
[00:12:06] Speaker B: Yes, 100%. Yeah. The basics are there and this is good news and bad news. Right. The good news is that this is a known thing. People have been doing knowledge management for a long time.
And the bad news is that you actually have to do it, which is the big challenge. And I do think more tools will be coming to help knowledge managers, to help them structure, to help them onboard content, to help them do that data analysis, what I sometimes think of as back office knowledge management work. I do think tools are coming for that to help that speed up efficiency, but it doesn't change the core needs, which is structured, readable, searchable, good content.
[00:12:56] Speaker A: Okay, so we'll watch out for some more products coming from nice, perhaps. But one of the things that nice likes to talk about is knowledge management. Today really being about AI management, can you explain a little bit more about what that means?
[00:13:13] Speaker B: Yeah, it's a good question. So when I think about AI management, I think about controlling an AI system to meet the needs of your brand and to make sure that it meets your brand promise. It's acting on your behalf.
And knowledge management can help you elevate that by feeding it really great knowledge about how your brand works. What answers to questions are how to help your agents, how to help your clients. So when you have a really strong knowledge management practice, I feel like I've just said this over and over again. Structured content meets the needs of your users, readable. All of that gets fed into an AI system and it helps manage that AI system in order to meet the needs of your brand and your users and your customers.
[00:14:04] Speaker A: I mean, you may be repeating it, but that, it bears repeating, right? I mean, it really boils down to just that. Yeah, well, maybe not just that, but certainly that's the core and the AI.
[00:14:18] Speaker B: Management knowledge management piece. I'll just say one additional thing, which is that many of us have heard the term or the adage garbage in, garbage out. Right? You can have the fanciest AI system in the whole world, but if you feed it bad knowledge, or poorly structured knowledge, or confusing knowledge from your knowledge management, knowledge management system, it's going to spit out confusing answers and produce confusing experiences for your users. But if you have that strong good knowledge, the good stuff, the not garbage that you feed in, you're going to get good experiences out.
[00:14:55] Speaker A: So if you talk about hallucination, it's hallucination because the data has gone in. The data that's gone in has been bad.
[00:15:04] Speaker B: Yeah, yeah. There's all kinds of reasons that, all kinds of things that can cause hallucination, and one of them is you've fed it confusing information.
Maybe. Let's say you have two articles from two totally separate wikis that have the two, two different answers to the same question.
The AI might spit out what you think is a wrong answer based off of what you know to be true. But it now is looking at two different articles that say two completely different things, and it's just a robot. It doesn't know how to decide what's true and what's not. So what you experience as a hallucination is bad data.
[00:15:44] Speaker A: I think.
I spoke with Aaron Rice with nice earlier, and he talked about you need to have data accuracy, but you also, you need to start with data fidelity. Right. You have to have faith in your data. So.
[00:16:02] Speaker B: Yeah, absolutely. Yeah.
[00:16:05] Speaker A: Okay.
[00:16:05] Speaker B: Did he give you his smoothie analogy talking about broccoli?
[00:16:11] Speaker A: I don't think he used the smoothie analogy.
[00:16:14] Speaker B: Okay. What's that same thing? Building an AI system is like making a smoothie, right? So your AI is just your blender. It's just the tool.
And if you are making a smoothie with bad ingredients, with, like, random stuff from the back of your fridge, or you're making your AI smoothie, I don't.
[00:16:33] Speaker A: Have a picture in my mind. Whitney.
[00:16:37] Speaker B: If you're making your AI smoothie from your just sort of forlorn wiki that you don't pay any attention to, you're going to get a gross back of the fridge smoothie. But if you make it with the good stuff, the strawberries and the pineapple at the front of your fridge, and you really know what's going into it, you make it with knowledge that you've seen recently, that you trust, that you care about, that you know is accurate, that is useful. Your knowledge smoothie is going to be delicious.
[00:17:05] Speaker A: I like that. I like that analogy. And leave it to Erin to come up with it.
[00:17:09] Speaker B: Yes. Yeah.
[00:17:10] Speaker A: Classic Erin Whitney. Just a few last questions. What trends are you most watching around AI and generative AI data knowledge management? And then what's your vision of where all this is going to lead?
[00:17:23] Speaker B: Yeah, I've got two trends that I'm super excited about. The first one is interconnectedness. And the way that I think we're going to see these AI systems start to be able to talk to multiple input systems, talk to each other. We saw that in Barack's presentation at interactions with the way he was talking about the AI, being able to remember when you've had a touch point with a brand, and every time you've touched the brand, every time you've talked to the brand. And then we also saw it at the Apple developer conference. I don't know if you watched that. I was glued to it.
[00:17:57] Speaker A: No, I did not touch that but I did interactions. And just so those in our audience who aren't familiar with interactions, that's Nice's customer event, and it took place in mid June in Vegas.
[00:18:12] Speaker B: Yeah, yeah. And the Apple developer event, same timeframe. They showed sort of a souped up Siri that can look at your texts and parse out restaurant reservation information just from your texts and then can take that with the maps application and say, okay, if you want to make your reservation, you need to leave at this time based on traffic. And so you're seeing this kind of like the promise of AI and the promise of these little pocket assistants starting very slowly to come to fruition, which is really cool.
And then in knowledge management, I touched on this a little bit. I am very excited to see the tools that help knowledge managers and that help structure content, that help you. Generative AI obviously can help you write content, but that can help you format it for different audiences, make it readable for who needs it to be readable, to do some of that data analysis, to help you structure and create the content that your users need or your customers need.
I think we're going to start to see some of that. And that's not a product announcement, just for anyone who's listening, any of our customers, that's not a product announcement. It's just a trend that I'm seeing as I watch the knowledge management space. I think we're going to start to see much more of that.
[00:19:39] Speaker A: Okay, so what's your sort of ultimate vision of where all this is going to lead? Whitney?
[00:19:44] Speaker B: I'm such an optimist. I think this is going to be great. I think we're going to end up with tools that leverage the best of AI. So their analytics, their data crunching ability, their ability to pull signal out of noise, and then present that back to people to enable people to do what they do best, which is be creative, problem solve, have empathy. And so I really hope we see more tools that sort of bring the best of both worlds together to help you sort of to be your knowledge copilot, right? To help you in the whole world of knowledge management.
[00:20:21] Speaker A: Excellent.
[00:20:21] Speaker B: Excellent.
[00:20:22] Speaker A: Okay. Okay. Any resources you'd recommend to people looking to better understand the state of AI and knowledge management and data driven cx today?
[00:20:30] Speaker B: Yeah, I think I get most of my news from podcasts just like this one these days. So to stay up to date with AI in general, I like the Hard Fork podcast. It's a New York Times podcast, just about tech in general. And then the big tech podcast, I also really like just to stay up to date on what all the technology players are doing for new folks who are new to knowledge management. I'll plug something that my team wrote. We wrote an ebook called Knowledge Rocks, formed after the what? I'm trying to make a reference and it's just off the, it's off the tip of my tongue, but it's all about knowledge management and how to get started. And then I like the KM world blog to keep up with knowledge management in general.
[00:21:21] Speaker A: Excellent. Well, those are some great resources. I'll check them out myself as well. And that is a great place to leave off. So thank you again for sharing with us, Whitney. Until next time, everybody. Take care.
[00:21:34] Speaker B: Thanks, Beth. Thanks everyone.
