AI and Product Marketing: Redefining Strategic Horizons

Julien Redelsperger : « And I'm super happy today to welcome John Rau. He's the director of product marketing at Visier, which is a Canadian company that provides people analytics solutions that include generative AI, of course, to focus on retention, hire the right people, and make smarter HR decisions. Today, we are diving into the intersection of AI and product marketing. Thank you for joining me. How are you, John? »

John Rau : « I'm good, Julien. It's good to be here. Thanks for having me. »

Julien Redelsperger : « Yeah, sure. That's my pleasure. Thank you for being my guest today. So, John, before we get started, what can you tell me about Visier? »

John Rau : « Yeah, great question. So Visier is actually a company that I didn't know a whole lot about until I got hired. It was very popular in the HR space. So if you think about these really big companies that have lots of data and lots of data about their people, they have a lot of systems in place to analyze, say, their financial data, to analyze their inventory, all these other parts of their business. And what Visier is, is it's a way to take all the amazing stuff that's going on with your people that exists in all these different systems, bring it all together, and give your management, your HR teams sort of like a 10,000-foot view of what's driving performance within your organization, what's driving resignations, and answer all these really, really, really cool questions about your people, not just on a macro level, but also allowing you to sort of like drill down into it. So if you think about it, it's people analytics, analytics for the people in your business. »

Julien Redelsperger : « And so what can you do with this data? Like, what's the output? »

John Rau : « Yeah, so a lot of people use Visier to understand drivers of performance, for example. So you might have a group, a cohort of really high performing employees. And as a company, you might want to replicate that across your business. So you might pull data from your applicant tracking system that talks about where they came from before or where they're located. You might take data from your engagement system that surveys them throughout their tenure, your learning management system that shows what courses they've taken, all these systems that house information about your people in very disparate, discrete places, that you can only get those insights, for example, about performance if you tie all that together and see what's actually driving something like performance with the data from all those different systems. »

Julien Redelsperger : « Okay. And so you are a product marketer at Visier. So what is your role as a product marketer? And what do you do exactly? »

John Rau : « You're right. I'm in product marketing. There are lots of product marketers at Visier. So I'm really lucky to be surrounded by a really talented group of colleagues who we always challenge each other and learn from each other. My specific background, and we can talk about this more later, is technical product marketing. So I have a bit of a developer background myself. And what Visier is really starting to do is look at how do we help people who are looking to purchase Visier or bring it into their organization better understand how it works with the rest of their ecosystem, how it fits into everything else, how they might set it up, all those things. And then for people that really want to take things to the next level, how can they take the technology underneath Visier, like our platform, customize it into something that really solves a problem that we at Visier haven't even thought of yet. »

Julien Redelsperger : « Okay. So there are several product marketers at Visier. That means there is not just one product marketer for one product. Is it like per feature? Is it per capability? Like how are you organized and how do you decide what's your scope in terms of product marketing? »

John Rau : « Yeah, that's an amazing question. That's a really good question, Julien, because at different organizations, I've seen different setups. So in some organizations, you might have a product marketer by feature or a product marketer by product line or a product marketer by vertical or a product marketer by audience. So at Visier, we have one product marketer that's focused around a certain set of products. We have another one that's focused on sort of a core product in our Gen AI product. We have another one that's focused on our technical buyers and our direct to business side. And then there's me who's focused on the platform. So it's not necessarily like a clean cut as well. It's not necessarily like a clean cut, like each person is responsible for an audience or a product. It's really through our management sort of understanding of the needs of the business and the audiences trying to find that perfect mix. And what I found at other organizations too is it's never as clean cut as you might expect. And that's just how we've sort of organized it here. »

Julien Redelsperger : « Okay. And so you started your career as an entrepreneur, as a developer, and then you switched to a marketing position. So why is that? And what motivated your transition from web dev to marketing? And how difficult or how smooth was it? »

John Rau : « Yeah, so that's a really good question, Julien. And I should mention, it's something that we're seeing a lot more in the marketplace now too. Well, when I say a lot more, maybe not as much as that sounds like, but someone reached out to me out of the blue on LinkedIn last week, and they sent me a stat, a statistic that they had pulled around developers or engineers and the percentage of marketing roles that hired developers or engineers. So I think it was, and I'm probably going to get this stat wrong, but I think a couple years ago, it was hovering around 2% of marketing hires had an engineering or development background. And now in 2024, that number is either 3% or 4%. So it's jumped something like 50% to 100%. The reason I say that is I think companies are starting to realize that if they want to have credible marketing to technical audiences who are very scrupulous, and to be quite frank, they don't actually like marketing in its traditional form, they're hiring people with these technical backgrounds to really make sure there's rigor in that messaging so that it resonates with the audience. So going back to my story, and the other reason I mentioned that is if you are in engineering or development roles, and you're curious about marketing, you can reach out to me on LinkedIn, and I'm happy to talk you through my journey and give advice in a bit more detail. Going back to my journey specifically, I graduated university with a degree in information technology, which was really cool. And I wanted to do something a bit entrepreneurial out of school. So what I did was I started doing freelance web development, something that I knew something that I did during high school, I thought it was like an easy way to make money. And I just kind of enjoy that kind of stuff. As I was doing that, I found there was like a demand for it. This was maybe in the early 2010s. So I was able to sort of hire a few friends, we call ourselves like a small agency, it was kind of fun. And then we started getting customers. And what we found was we would get the customers, they'd pass through the website. And that was all we had like within our wheelhouse. So then that relationship would end, it would be done. But then what we realized when they were trying to get a website to the questions that they were asking and getting a better understanding of sort of the landscape out there and how we can improve the business to offer like ongoing value to them over time, which helps our business as well. We realized they weren't actually after a website, they were after business, new business using digital means. So that's when we became more of like a full service digital marketing agency, where it was like, yeah, we'll do the website. But also like, let's talk about AdWords, your engagement strategy on social media, how you're measuring all this stuff, marketing automation. So that's how I got into marketing from development. And then from there, I moved into a more generalist marketing role at a company here in Ottawa. And then a few years had passed that and then I guess the last company I was at, TinyMCE, they're like an embeddable software component. They reached out to me on LinkedIn and said, "Hey, like we see you have this technical background and marketing background, you want to come work for us?" So that for a couple of years, and then spoke with some folks at Visier, similar story. And here I am. »

Julien Redelsperger : « Okay. So would you say it is easier to have like a tech background and to learn marketing or to have a marketing background and to learn tech if you want to be a successful product marketer? »

John Rau : « That's a great question. And maybe it's good to clarify what product marketing is right now. I'll just go through like my spiel just because some of the folks in the audience, they may think it's just marketing a product and that's part of it. But to give a bit of background of what product marketing is, you're really trying to understand A, what the product is, B, who your target market is, who exists in that, what their pain points are, what the challenges are, what the alternatives are, the language that they use. And then you're trying to match up what your product offering is with that audience or those multiple audiences in ways that resonate with their world versus you talking a lot about your product. So you're contextualizing everything there. And the really cool thing about product marketing is as you get to know that audience on the outside that you're trying to tailor your product and your messaging to, you develop these really cool market insights that you can then bring back to your product and engineering teams and help to influence the direction of the product. So that's what product marketing is as a whole. Back to your question, it was, is it easier to learn technology stuff and then go into marketing or learn marketing stuff and then go into technology? I don't actually know the answer to that question, but I would say with the advent of, this is a great lead into the podcast, the advent of things like generative AI and just the proliferation of low code, no code tools and ways to learn in so many different ways, it has never been a better time to learn something new. The crazy thing is like, you don't actually need a course to learn something. You don't need to pay someone. And this is not good news for course creators. And I, it makes me sad, but you can just go into ChatGPT and say, Hey, I want to learn this topic. Walk me through everything I need to know. Give me a quiz. Give me an exercise to do, and it'll probably do a pretty good job. So I would say if you're, if you're interested, if you're a non-technology person and you're curious, just dive in, come up with a project, start learning something, build something, especially if, if, if you want to be in product marketing, a great tip is just become your audience. Like who are you marketing to study the stuff that they're studying, try and accomplish the tasks that they're accomplishing. And then you'll naturally be able to communicate with them in a way that makes sense to them. »

Julien Redelsperger : « Sure. Yeah, no, absolutely. No, that's great advice. So you, why are you interested in generative AI or AI before it went mainstream, like about a year and a half ago? What did you know about AI then? And what do you know about AI now? »

John Rau : « Wow. So much has happened in the past, in the past while. And it really, it's really just changed so much, I guess, before chat GPT, like hit the stage in maybe it was February, 2023. So about a year ago, and I know that's not when it hit the stage, but when it really gained in popularity and momentum I always thought of AI just as like a sci-fi thing. There had been so many AI startups, so many false promises that never got delivered on all this hype around this technology over the past, like 15 years, let's say that I just sort of, anytime I see anything related to AI, I just be like, Oh, okay, whatever. Like call me when it can actually do something useful. And then I went on a paternity leave with my wife and our one-year-old daughter. We went to New Zealand for a month. I was kind of tuned out, just hiking through some mountains and having fun with my family. And then I came back and I was like, everything was talking about chat GPT, chat GPT, what the heck is this thing? And then I was a bit hesitant because I was like, Oh, this is another stupid AI thing. I waited about a month and then I started playing around with it. And I was like, wow, this can completely transform how we do work as a whole, but specifically myself, how I can improve how I do things and focus on the work that I love and pretty much like outsource the boring stuff to someone else. »

Julien Redelsperger : « Okay. And so what do you think changed in your role as a product marketer since chat GPT, since generative AI and how do you use it on a daily basis if you do use it? »

John Rau : « Yeah. So to answer your question, I definitely use it. How it's changed how I do my day-to-day job, I'm still doing the same stuff. I'm still focusing on the same problem, solving the same problem, working on messaging, understanding the customer, matching the product with the customer needs. However, I would say I'm probably working or at least outputting at two times the rate and I'm not doing the stuff that I don't enjoy the boring stuff of product marketing. And I'll, I'll, I'll double click into that scouring, um, like research for things. Like for example, if I'm looking for some trends, the first place I'll go is chat GPT and then I'll validate that through Google or something like that. Um, summarizing content, summarizing documents, getting started with an idea and outline for an argument, um, getting a first pass of a review on something. These are all ways that I use chat GPT. I use the premium one chat GPT for well worth whatever it costs. I think it's like 20, 25 bucks. I use it for stuff like that. So some examples, um, I, you know, because I was trying to get ready for our podcast, um, I scoured through my chat GPT for like history and it was funny. I actually used, I took screenshots of it that I use chat GPT. I pumped them all into chat GPT for that. I said, turn these screenshots into a list and then it did like OCR on it and turn it into a list. And I said, that list is too long. Cut out the non work related stuff. Like me asking about like what a toddler can, like a toddler can eat like an apple peel or something like that's not going to be useful for our conversation today. Fine. Cut out all that. Okay. It was still like 120 things that I said, categorize it. And then it categorized it all for me. So anyways, that's not even me getting into how I use it for work in terms of product marketing. Um, the way that I use it is if I have an idea of trying to accomplish a task, I'll start off like, Hey, this is like what I'm trying to do. Can you help me put together an outline or a way to approach this problem? So I want, let's just say, um, great example. I want to launch a new product in market X. Here's a rough outline of what the product does, who the audience is. How would you approach this problem? Now it's not going to do all the work for me, but it's going to give me a pretty darn good starting point to have all the things I need to think about and, and, you know, explore. And I can say, ah, I'm not going to worry about that, but I've got a great outline. Great. So I take that, start building it out on my own, whatever. And then, uh, sort of every once in a while, put different sections back in for feedback. What do you think of this? Um, what might someone in this persona think of this? If they read it with what criticisms might they have? How might they counter this? What, uh, objections might they have to this sort of a proposal to them? And then again, it doesn't get you a hundred percent of the way there, but it can speed up that process of thinking through all of these things. Um, I think that's a really important one, like exposing weaknesses in an argument or, you know, testing out how different people might interpret a certain message, something like that. Um, and then once I have like a final product, I'll put it back in there and I'll get some feedback on it. Um, even like scenario simulations, you can use it for that. Uh, I mentioned earlier, learning new skills, all that stuff. Very, very helpful. Um, the only thing that, you know, the only tip, and I'm sure everybody on your, your, on the podcast would probably mention this is you do have to be careful about what you put in there. Don't be putting like your whole business strategy into chat GPT, because it's training on everything you put in there. »

Julien Redelsperger : « My next question actually, John, is how do you manage data privacy? And do you use chat GPT on your own initiatives or is it like encourage, uh, like within your, your, your organization? Do you need, I don't know, permission to use chat GPT at work? Like how, how does, how does it work? »

John Rau : « Yeah, it's a really good question. So, um, our policy at Visier is you never use anything related to customer stuff, customer data, customer names, anything like that. So that's like a big no, no, we're dealing with very sensitive info and we are not touching any of that stuff that will never make it into any, any model, any sort of names or like that just doesn't. And especially in my job as product marketing, I don't even have access to that stuff to be honest. So that's like a big thing that we, we care a lot about and we take very seriously across all of our initiatives. For a product marketing perspective, luckily marketing is, uh, it's important. Your marketing strategy is really important. You don't want to expose it all to a large language model. So a lot of things that I'll do is actually have in chat GPT, if you have the premium version, you can create multiple copies of chat GPT and give it instructions of like who you are, what you're trying to accomplish. So I have my generic chat GPT that knows nothing about me. Like it, it doesn't have any instructions. So what I'll use that one for is if, if I just, I'll, I'll take what I'm trying to do, take any specificity out of it that could identify it to our company or even to our, our industry even, uh, and just sort of spit some stuff in there and, uh, and see what happens then if I do need to say like, Oh, okay. Like the strategy I'm working on, it's targeting, uh, HR for example. Um, I'll say that, but very, very careful about what we put in, uh, and what I put in, um, because privacy and our, our strategy, that's like of, of top importance to us. »

Julien Redelsperger : « And do you know if your colleagues, the other product marketers are using chat GPT as well? Do you, do you talk about it? »

John Rau : « Oh, that's a great question. Yeah. It's, it's, uh, no, it's, it's, it's really interesting, right? Because, um, especially when it first came out, I know a lot of people were afraid of, um, saying that they're using it cause they almost felt like they were cheating or something, you know, like it's like our approach and the team that I'm on, our approach is definitely, uh, if you, if there's a tool that's going to help you do your job better, use it, you know, use it responsibly and, uh, and, and share your learnings. Like we, we talk about it often, how, how we're using it and what it's helping us do. »

Julien Redelsperger : « How do you think generative AI will change your job in the future? Like in five years, 10 years from now, do you think product marketer will still be a thing or would it be, I don't know, outsourced and, and, and, and everything would be like organized and created by AI? »

John Rau : « In any profession, you almost have a responsibility to understand this technology or any emerging technology and understand how you can leverage it to the advantage of the business that you're working for your team and yourself as an individual, the closer you are to a new technology, the less likely it is to make you as an individual obsolete because you're learning how to leverage it. You know, it's strengths, you know, it's weaknesses. You can talk about it to management in an educated way. So that's my first point. Uh, and I plan on staying close to Jenny. I find it fast. I, if anything, it's just really interesting. This the question was around what would product marketing look like five, 10 years ago, five, 10 years in the future. So I don't know, but I do know that if as product marketers, we're close to this technology and we're constantly using it to our advantage, we will hopefully find new ways to use it to make product marketing and even more impactful function on the business rather than make it something that's automatable. Because I also don't want to live in a future where we have a bunch of companies that are managed by AI with a bunch of AI bots competing with each other. Like that just seems very like dystopian and post-apocalyptic. And I just can't, I do not want to live in that world and I'll do everything I can to avoid that kind of world. But at the same time, it's a tool. Let's use it. »

Julien Redelsperger : « Okay. And how do you, how do you train yourself on AI? Because it is constantly evolving. You are just mentioning at the beginning of the podcast that you left for a month traveling. And when you get back, like everything changed. How do you keep up your marketing strategy, but also your personal information about like what's new on AI? Because sometimes I feel we all have a little bit of fear of missing out because every week, every day, sometimes you have new features, new capability like that. It could be overwhelming sometimes. How do you deal with that? »

John Rau : « So you're right. It can be overwhelming sometimes, especially if you're new and you're just sort of getting into it and you don't know where to start. Like put it this way, you sign up for something like chat GPT and it's just an empty box. What do I type? People are telling me, I mean, I need to use this. Where do I even start? Right? So that's super intimidating to begin with. And then if you have the wrong output or you have one bad experience, you might just be like, oh, this isn't for me. And you'll give up on something that could be incredibly beneficial to you. So how do you stay up to date? How do I stay up to date? I cannot keep up with all the evolutions in Gen AI, nor should I, because there's so much going on in so many different spaces. There's image generation. It looks cool. I don't care about it. It doesn't help me with my job. There's a video content, a brand new thing. This stuff, I know I work in marketing and I should maybe care a bit more, but my job is product marketing. So I'm trying to stick to my lane and think about how can I use this technology to best help myself as a professional, my team, and the company that I work for achieve our aspirations. So by staying focused on how I can use it to help accelerate product marketing, get the most out of our product marketing function, it gives me a bit more of a scope that I can work within and I can then figure out what I want to experiment with, narrow in on tools, what I see like a new tool every day. It's very easy to just say, nope, keep going on with my day versus trying to keep up to date with every single possible thing. So that's how I do it. It works for me, might not work for everybody else, but I don't find myself overwhelmed with new AI features because I'm very good or I've gotten very good at choosing what to pay attention to and what not to. So rule number one would be stay focused. And rule number two would be which channel do you use when you want to know more about AI? Are you, I don't know, listening to podcasts, reading books, LinkedIn? »

John Rau : « And it's funny, I don't actually follow AI thought leaders. I follow thought leaders in marketing and I follow thought leaders in the HR tech space that we operate in at Visier. And what I've found is I'm now getting thought leaders in those spaces talk about how Gen AI and AI help them in their role. So it's contextualized for me already. I'm not like a big social media guy. I have Twitter. I can't say that I use it that much. Facebook, barely use it. Instagram, I think I have it. I couldn't tell you. LinkedIn, that's like sort of the one platform for me that I log into. That's where I consume a lot of my work and consequently AI related content. »

Julien Redelsperger : « So I know you don't work closely with your clients, but you work in the HR and tech ecosystem. What do you think HR people think about AI, generative AI, actually? Because at Visier, you have a generative AI component into your platform. So what do your clients think about it and how does that work? »

John Rau : « Well, I've heard some clients and customers talk about it and they're genuinely curious about it. They want to know how can we use this technology to better help us as a team, better help our managers make really good decisions for the business. »

Julien Redelsperger : « And could you give me an example on the type of like use case with Gen AI with Visier? »

John Rau : « If you think about a traditional organization and what does that even mean, traditional organization, but a very, very large company, thousands of employees, it can become really challenging to know what people are up to. For example, like the example we used before, like what is driving really high performance? What are some factors that are impacting low performance? Is it the data that's coming in? It's not anyone's fault, it's just a very manual process sometimes. Say like the CEO asks, "I want to know what's impacting performance in Singapore for our sales team?" Okay, wow. So the HR team then has to unpack what is that question that they're asking, then they'll work with some analysts who have to find out where that data might be stored, pull it all together in a unified view, come up with a dashboard, share it back to the CEO maybe a couple weeks later, and then the CEO's like, "No, that's not what I was asking." So then they got to go back to the drawing board. It can just take a lot of time and business leaders want insights today when they ask the questions. So then if you think about what about if there was a world where rather than a business leader or a manager having to ask a team to generate some reports to answer a question and have that time delay, and maybe the business leader doesn't know if they're asking the question in the right way or the question's being misinterpreted or something like that. What if there was a way for them to log into their people analytics solution, start a chat with a Gen AI chatbot and say, "Tell me information about what's driving our top performers in Singapore." And then the chatbot might be, "Well, what do you mean by that? Are you asking this, this, or this? Are you talking about sales performance? Okay, let me pull a chart." And then the chart pulls up and that happens almost instantly. It knows where to get all the data, it's looking for all of it. It pulls it for that business user. That business user doesn't need to be a data analyst or an HR person to then be able to interpret that information. That's what we're launching at Visier. It's V, it's our generative AI solution that goes alongside with all this people data that these companies already have so that those managers can get those insights as they're thinking. »

Julien Redelsperger : « Okay, cool. That's interesting because I think one of the important thing about AI and Gen AI is how to ask the question. Because it's not because you have the data that you can get the right output. And I read multiple articles and blog posts about prompt engineering, which could be the great new skills to acquire for a marketer. What do you think about it? And how do you work your prompt engineering skills on your own? »

John Rau : « You know, to someone new to generative AI, they may see the word prompt engineering and the word engineer and think, "Whoa, man, this is not for me. This is like too technical or something like that." And it really isn't. I mean, it's a term that someone came up with and it's stuck. At the end of the day, all you're doing with prompt engineering without even knowing it is you're just learning how to ask the AI engine what you're looking for. And that comes with practice. It comes with seeing what sort of comes out. It comes with iteration and experimentation. That's how I use it. I don't call myself a prompt engineer. I don't have like that credential on my LinkedIn. But what I do do is I try to give a bit of context in my first input into chat GPT, what I'm doing, product marketing. And then I see what comes back. If it looks like it needs more guidance, I correct it. I say, "Well, actually, I'm looking for this." And I just keep going. It's a conversation, right? Chat GPT is a conversational chatbot. One thing I will say is I try not to over prime it at the beginning because that can be really inefficient because it means I can invest a lot of time initially trying to give it background information that may not even work. So what I might say is, "Hey, look, I'm hoping to accomplish X, Y, and Z. Here's some initial information. What else do you need for me to start generating an outline or start providing viewpoints on this topic?" And then it'll start asking me questions kind of thing. So then it sort of like turns it around on chat GPT to tell you what it wants to achieve the outcome that you're looking for. But yeah, prompt engineering, don't get too intimidated by it. See what's on the internet, but also recognize like anyone can call themselves a prompt engineer. Anyone can charge $100 for a PDF that contains a bunch of pre-made prompts. Look for inspiration. Come up with what works best for you.

Julien Redelsperger : « I used to speak to chat GPT the same way I would speak to my four-year-old. It's like simple questions, one task at a time. Rephrase, rephrase again, re-inform. And it usually works. »

John Rau : « Well, it's really interesting you say that, Julien, because it also has a memory. And I am going to probably not get this exactly right when I describe it, but this is close to how it works or close to how it worked when I looked at this a few months ago. So there's a certain number. Okay, actually, let's go back even further. Chat GPT is just like the front end of OpenAI's generative AI engine. So it's just like user interface and they've built a nice UI around it and some logic for sending and receiving queries to OpenAI. My understanding, and again, like I said, full disclaimer, I could be wrong with about 50% of this, but I'm probably 50% right too, is it can only send up to 4,000 or 8,000, they're called tokens, which are like fractions of a word, to OpenAI at a time. OpenAI analyzes each of those requests discreetly from each one. But what chat GPT does is when you send a request or send a chat through chat GPT and there's a conversation history, it's actually summarizing key parts of that conversation and sending it along with that so that the AI engine can send back something. Now why that's important is if you're having a really long conversation with chat GPT, which I've had some very long conversations with it, it will start forgetting stuff that you mentioned at the beginning. So if you can start seeing things that seem a bit wonky, you might have to remind it of some of the key facts or the key points that you already discussed with it to get it sort of like back on track. Or maybe start a new chat. It could be worth it as well. Yeah, there's lots you can do, but just be aware it's not perfect and it doesn't remember everything. »

Julien Redelsperger : « John, one last question for you. What advice would you like to give to someone who would like to become a product marketer in 2024? »

John Rau : « It's a great question. I actually get asked it by a lot of people because they find product marketing very intriguing and very interesting. My advice would be, even if you're in a job where your job title is not product marketing, you're not even on the marketing team, get to know folks at your company who are in that role and try to work on projects with them and try to understand what they're doing. So you can still, as someone maybe who's in developer relations, you can still pair up with a product marketer on a project and start learning how they do things and how they approach things and understanding what templates they use, that kind of stuff. So that's number one. Number two, just go on LinkedIn or wherever you consume your content and start following product marketing folks. They post a lot of really interesting stuff. But number three, talk to ChatGPT. Ask it, if you're working on a project that has a forward facing thing, something that's outside facing, ask, from a product marketing perspective, how would you make this even better? And then it can start teaching you different tools and techniques and ways to become a really good product marketer. And just reach out to me on LinkedIn if you want to chat. »

Julien Redelsperger : « And so John, how do you talk to people when you work remotely? I know it could be a challenge sometimes, specifically for new employees. Any advice you want to share? »

John Rau : « Yeah, it's hard. On one hand, I love working from home because I get to see my daughter after daycare. I get to bring her to daycare and we spend a lot of family time together, which is really, really, really nice. However, at the same time, it can be challenging when you're working in a bedroom in your house and the workday's over to just go straight from really intense work to going and hanging with your family. So that can be hard. And then to your point around connecting with your coworkers and stuff like that, you have all these Zoom meetings scheduled and you're trying to be respectful of everybody's time and really trying to collaborate over these digital channels. It's very easy for messages to get lost or for things to get misinterpreted. So I think it's really important to almost overemphasize body language, overemphasize a message, follow up with people if you feel like things didn't go exactly as you had maybe hoped and just clarify your intentions because everybody's working together to support each other and we want to make sure that everybody feels like they're part of a team. It takes a bit of extra effort than working in person. »

Julien Redelsperger : « Yeah, I guess don't be shy, specifically for introvert people. »

John Rau : « Yeah, that's a great tip too. »

Julien Redelsperger : « Okay, cool. Perfect. Well, that's a wrap on. So at the end of each episode, the guest must answer a question posed by the previous guest. After that, you'll have the opportunity to ask a question for the next guest. Hi, are you ready? »

John Rau : « I'm ready. »

Julien Redelsperger : « Perfect. So here's your question, courtesy of Andrew Just, who is the founder of Train Fitness, a Toronto-based startup that develops an AI-powered fitness application. We can listen to his question right now.

Andrew Just : « Are all of these startups and companies that are building AI this or AI that, or basically wrappers around chat GPT, are those real companies? Are they going to be around in 10 years or are we in a bubble and we're going to see similar to what we saw in crypto, a huge rush and then a huge bust and then pop of that bubble as well? »

John Rau : « Andrew, great question. That's a really, really good one. So I do have an opinion on this. I think, and the reason I have an opinion on this, because last October I was in San Jose and I did a talk at AI Dev World. And the topic of my talk was, don't just add AI to your app, make it useful. And one of my arguments to Andrew's question was, it was funny, when chat GPT first became popular in February, 2023, I think it was, the apps that said that they had AI on them, everybody wanted to use. But then everybody said that they have AI in their app and then it just became noise to everyone. That's maybe in the spring. And so it's now, AI is no longer something that companies necessarily need to lead with, but it's more like going back to what a company has been doing that they've been really, really good at. Don't just say you have AI. How does Gen AI actually help your existing and future customer base achieve the same outcomes that you were doing for them, but even better, faster and more productively? Those I think are the companies that are going to make it in the long run. I think there's a lot of really cool wrappers around Gen AI and chat GPT, but you can actually do that yourself in chat GPT now, if you pay like 25 bucks a month, you can make your own GPT. So I don't know if that's really like a long lived, sustained momentum thing. Is it like crypto? I don't know if that's quite the comparison I would make. It's a good thing. It's a good example, but I don't think like, there's definitely value there. And it's a question of not just slapping it on your app, slapping it on your homepage, but using it to add value to your customers day to day, make their lives easier, help them achieve outcomes they never thought possible. And those are the AI companies that will still be around five, 10 years from now. That's my prediction. »

Julien Redelsperger : « Cool. Perfect. Do you think it's smart to embed AI from like open AI or chat GPT with like proprietary LLM? Or do you think it would be best to have your own proprietary AI system to make sure there is no dependency on third party AI system? »

John Rau : « Yeah, it's a really good question, Julien. And if you don't mind, I'll provide a bit of an example of how we do it at Visier. I know we're sort of approaching time here, but I think it's really important to understand there's different ways to leverage an LLM. So if you look at open AI, or Google's one or cloud, or, you know, there's a bunch of really good ones out there. It's going to be really hard for an individual company that's not in that space to, first of all, train or hire a bunch of ML engineers, which are very expensive, as you know, and replicate that level of quality and natural language processing and all the benefits that come from that. So then it becomes the question of if you're not like, I don't know, Microsoft or Google, how do you leverage this technology in ways that can help your customers? So the way that we've done that at Visier, and I was talking about this with V, our Gen AI assistant that helps business managers, HR teams get insights quickly in natural language is we use one of those popular LLMs, right? It only acts as a translation layer between the query that people are asking, and all of the data that we have in this amazing data system, data store that we've been building over the past 10 years, all of our customers' data that they upload into Visier, which interestingly enough, as we were building Visier, we developed almost like a query language, kind of like SQL structured query language, but for Visier, that is well documented, that can pull insights from the data store. So the way that we use the LLM, I'll get back to answering your question, is people type in their query, the LLM tries to interpret what they're looking for, the LLM knows the structure of the data really well, the fact that there's an employee object that it's related to a requisition or whatever, but it knows nothing about the actual company's data, any of the records. So what it does is it takes that query, translates it into the Visier query language, and then that goes somewhere else, our systems, which are not training an LLM or anything like that, then perform that query, it's formatted into a chart and insight, and then it's delivered back to the user. So no customer data ever touches the LLM, which allows us to offer this capability in a really secure way. The reason that I bring that up is because there are ways to leverage this technology, and this is just one example, without worrying so much about exposing or training all your data. I think Visier's done a really, really good job of thinking up a really unique way to leverage it. »

Julien Redelsperger : « Cool. Perfect. Well, thank you for sharing that, John. That's great. So now, what question would you like to pose for the next guest? »

John Rau : « So what I'm really curious about, because I know a lot of teachers, and they have mixed opinions on generative AI. Some of them use it, some of them, their students use it, some don't use it, some don't let their students use it. Is AI a net positive for education, or is it a net negative, and why? »

Julien Redelsperger : « Oh, great question. Thank you so much, John. It's been an absolute pleasure speaking with you today. Thank you for joining me. »

John Rau : « Thank you, Julien. Good job. »