Hi everyone, I'm back. Did you miss me? I've really missed recording to be honest, and I can't wait to share this episode with you. It's about AI and product management and how product management is evolving towards the future. Who better to join me in this conversation than bond on Jolt Singh, product strategy expert and returning guest. So it was a real blast. Enjoy. There's so much product content out there. It was there last year as well.
And now even more that the basics of product, right. And the sort of the, I call it the cult thinking about product, like everyone just knows Marty Kagan and all these big authors, those basics have not gone away. I mean, you still need to think about the customer, think about need, still need to think about the customer problem. That doesn't go away. Yeah, only the way you solve it becomes a bit different.
So I when I start looking into AI for product and this happened about 6 to 9 months back when mostly all companies start looking into OK, how they how they can best utilize it. Yeah and AI has been around for a long time as we know it, right? Like the machine learning models and predictive analysis has been there for a decade.
It's only, I think the ChatGPT that sort of, you know, got people into large language models and, you know, generative AI. That's that's what basically caught up. So I figured out that when I was writing more about these advanced topics, how to apply AI in product, that got more traction and people were willing to share it more. So there's always a lot of word
of mouth. So people want to tell others about, hey, I did this great article about AI and how to apply it and I really see a value in it. So I saw a lot of these metrics change for my newsletter in terms of word of mouth, more sharing, more audience and more open rate of emails. So all these good metrics went up, but it's also something that I might, I'm not an AI expert. And that's the dilemma you always have as a writer, right?
That how much do you ramp up so that you can give what your audience wants? Or should you just keep writing what you know? And, and you always have limits, right? You can't just keep writing if you're just writing from what you know, you'll always read some sort of a threshold. Yeah, yeah, you get that, yeah. Then specifically for AI, did you go really in depth with regards to OK, how does this work more on a technical level or more from a usage standpoint, Where is it affected?
Yeah. I think when you think about product and the product has so many dimensions to it, it's cross functional, right. I mean, if you talk about organization, you require probably a large part of the organization to contribute in making a product successful. That goes about marketing, operations, engineering, design, right. So it's, it has a cross functional nature to it. At the same time, what I try to do a little bit more is to, to figure it out at 2 levels.
So one is the product life cycle. So different stages of product and how AI can help in those product life cycle. And the second dimension is productivity, right? And that's productivity for myself or for the product manager itself, for example, productivity for the product team with design and engineers included, and then the productivity for wider organization. So if you just think about this on 2 axis, so there's productivity at this individual
team and organization level. On the other axis, you have, OK, which stage of life cycle are we talking about when we talk about productivity? Interesting. And and that's where, and then it just opens up so many elements about where you can embed AI. But I and we can talk about that, right? But that's where I see that people should think about anything about product. And yeah, I think these are the 2 dimensions people would think about.
Interesting. Yeah. I'm wondering if I'm missing out because I feel like last year or I think it was a year and a half when you and I spoke last time, I was still very much hands on full time software engineering. And then late last year, this opportunity came up in December of doing a product management role, specifically at ING that was also in the ESG domain. So the domain I really was enthusiastic about and I always wanted to try product.
And since I'm in a consultancy, I figured why not? My manager said, yeah, absolutely you can do it. So I got this role and I got this responsibility. But a lot of my like fundamental knowledge is more so from hands on software engineering experience. So I learned a lot this year. I feel like one of the things that I've noticed the most, I think is that there's a lot of people problems in trying to figure out how we collaborate
and cooperate. And sometimes there's even too many people that have an opinion about something and you have to educate them about, OK, how can we become really effective, really productive, right? That's even without any AI capability specifically. But this product that I have, it does have an AI component to it. Also the environment that we have AI still being kind of, I think tread it carefully in how
much we can adopt. So I haven't really dabbed with any kind of Productivity Tools for myself or what I do, but I do think it's fascinating. And that's specifically why, why I wonder if I'm missing out. Yeah. Are you using any AI in your day-to-day? Yeah. So maybe starting on that productivity dimension is let's start over there, right.
And for myself, I, I have evolved into a product leadership role now, right where I'm not, let's say, doing the the hands on work a lot of times, but I stay in touch because I manage product managers and also write about it. So I keep in touch. And what I found is that it just takes a bit of commitment, but you can find an AI tool for pretty much everything repeatable that you're trying to do. Yeah. Anything which has a repeatable element to even if it's complex, right?
And I can just take some examples. So my usage of AI and probably this is good for product leaders to understand because I, I, I think that this is a good thing for for them. I use AI for understanding the market research, discovery and eventually leading to some sort of a strategy as an outcome. Yeah, I actually hate today going to Google and searching for things because it just takes so much time as. Compared to, yeah. As compared to some AI tools out there and we can talk about it.
So let's say I'm doing market research about, OK, where's the competition going? Where's the market going? How's the economy going and what products make sense for the customers and stuff like that. Initially, I would spend probably at least 30 minutes, one hour just lingering around on Google websites and stuff and probably creating a document somewhere where I can write my findings.
Now I'm mostly I experiment with about two to three different GP TS or tools and I just keep experimenting which one is giving me a better answer to my needs. So I I, so writing prompts is really important and how you write prompts has become really an art 222 people with same tool may not get the same output. So that's the first thing. But for me, the, the, the biggest thing that AI or generative AI is solving for me is spending less time on research and product discovery work.
And it, it also shortens the time to communicate things to your team because you can, if you're using the right tools, the right prompts. I mean, obviously all those preconditions are there. I'm able to communicate much, much faster about what's happening to my team in
different context. That has been the biggest productivity shift for me to an extent that even in real time, like if you and I are in a meeting and I have a screen while we're talking, I can figure out some things on the go and Google. It was just impossible to do that right. If you're, if you're asking about, hey, but do you know that how will this go and that go? And I can just quickly get an answer in a few seconds. So I think it has just become my go to tool.
There's always something open on the side that I can just figure out. Aren't you scared about like the the challenges with regard to accuracy? And also sometimes when I'm researching something or looking into, let's say market research, I find myself being biased towards a certain angle. And then I might even write my prompts or I might not realize that I'm being also led by my own biases. With Google, I kind of see everything that's on the surface and I can deep dive.
I think that's the same challenge there. But like with generated content, it's like, yeah, you don't know where it comes from. Yeah, true. And I think that is where I, I'm conscious of that fact and do not rely on, let's say one tool. Gotcha. Mostly, yeah. And I think the biggest mindset shift anyone can do in product is just be open to experiment.
I don't think you should just see it as the final thing or that this is the tool that will now give me all the answers because of the reasons that you said, accuracy, hallucinations that generated wear does. So I always go in with a bit of experimentation mindset. If I have time, I'll just search on one tool and see. OK. Are they missing the recent information? Because maybe it's giving me information which is like 2022 because some of this generative has been trained on content that
was a bit dated. Yeah, ChatGPT is one of those tools which is slightly trained on an outdated content. So I don't rely on it for anything recent. I rely on it for like general business models and patterns and stuff. Yeah, which was, which is generally true and may not be true today. But then I moved to Google Gemini, for example, to understand things which are within my knowledge scope.
So let me explain it right. So the good thing about Google Gemini is that it can actually, it can train itself on your own. Google Workspace, for example. Let's say you're a company which only works on Google. And there are a lot of such companies like Google Docs and Google Sheets and Gmail and everything. Google Gemini can go through thousands of documents and just tell you what's happening. Or maybe you you're just looking for documentation within your company. It's not the open web.
And I found Google Gemini really trains itself well on your own Google Workspace or companies Google Workspace because Google has that ecosystem of tools which ChatGPT doesn't have. Very powerful. It's very powerful, right? Even to an extent I would this morning while coming here. I just checked with Google Gemini. Hey I think I have a calendar invite. Where should I drive to? And it gave me a Google Maps link. I did not even open that. E-mail for that. I love that.
Yeah. So, so because it is trained on your own sort of ecosystem of tools. Gotcha. And then thirdly, is the the tool I love the most is perplexity. OK, yeah, I've heard good things. Yeah, because it's the only tool that gives me source or the link from which it picked up the information. So the accuracy part, at least I can check the link even if it's not accurate, I can just go to the source and say, OK, where is it picked from? And secondly, it's trained on
recent data. So I can even ask them, OK, what's happening today in in Hilbert, Sam or in Amsterdam, right. So it can even tell me about things that are happening today or going to happen today. Yeah, so, and that's just three examples of you need to see which tool you want to pick up bases, what kind of problem you're trying to solve. So and you need to be open to experimentation and there's so many tools more out there right,
besides these three. Yeah, I think that's such a crucial skill, like that experimentation mindset, not just with AI, but also how you, for example, adopt new features in your product and see what sticks and what doesn't. And that mindset of letting go things that don't work and testing it with data, I think makes a lot more sense in that aspect. One of the things you noted was also it helps you with regards to communication.
And now for me, one year in this role, I've noticed that communication is like everything. If I see in a meeting that some people are saying something and I'm like, they're just painting part of the picture, I can jump in and kind of paint that picture fully because if we need buy in, we need to communicate it well. I feel like that's where I've been making most of my difference, also being kind of this person that has more technical knowledge so I can have those conversations as
well. But how have you been leveraging, let's say, AI tools for communication? Yeah, I think one part of it I would say is that there are a lot of elements to communication, right? Like before you can communicate async or sync, depending on, you know, how you want to communicate. You need to have certain artefacts ready before you can
communicate, right? I mean, there's oral communication, obviously, but where are you picking your sources of information, relating them into, let's say some sort of a document and then sharing it across to different set of stakeholders? That's async communication. And sync communication is when you're sort of giving the
context within the meeting. And I think for me at least, where AI makes a lot of differences, async communication, because as humans, we need to either be, let's say, you know, be reminded by some sort of an reminder or To Do List that, hey, we need to do XYZ. And communication is always about, OK, which stakeholders are less communicated over more communicated? And what, when was the last communication done? Can I do it again? Should I do it again?
All of those things. And that is where I, I feel that any kind of AI tools that is either trained on the information from which you pick up stuff. So it could be market information, your own company documents, your Confluence pages or whatever it is. I always like to use the same tools that are trained on top of the data that your company uses to basically extract information quickly and then communicate back to them. So it could be about OK for this product.
When we launched the first feature back into in 23, what were the top three bugs we found? Yeah. And this is such a complex thing to solve as a human because we'll go through some kind of a folder, go through some document, and then scroll and stuff. This can be done probably in 30 seconds if you have an AI trained on your own set of data and information. Beautiful.
Yeah. And lot of enterprise tools and I learned it from the enterprise tools like Atlassian with its Confluence and and JIRA is, is is able to do that now if you if you have that tool set and also a lot of other enterprise tools. So communication I think changes because now there are a lot of companies where, you know, some individuals have the the full knowledge about what's happening in the company because they just have been in too many projects. They've just been there, yeah.
They've just been there and you know that guy on that person or that Lady is so important for us because she or he have been there in this project and you know the brain, the brain dumps of the company, right? We all know them in the company. Like wallpaper? Yeah, yeah.
And, and I think that's what AI could solve a lot because if at least these people are writing things somewhere or documenting things somewhere, and that's, that's part of the data on which AI is trained, anyone can basically get the same set of information and through human chat interface, right? So you're not going through documents. You're just asking, hey, where can I find this and where, where do I go for that? So that that's sort of the
preparation for communication. So I took a step back. So how do you basically get the information to communicate the context? And I think we become as easy as just checking in, chatting with an interface to get the business context. Today, the business context only lies with the few individuals. Yeah. And then it will be for everyone to to know and just ask. That's incredible. That'll that'll be a big change.
Yeah, yeah, I love that as well. I've been thinking from a product perspective, but from a usage standpoint as well as from a product manager perspective. AI is going to be everywhere, and I feel like everyone's trying to find where it can be effective in their own tool, in their own suite.
I saw a demo with Myro and Myro. You have like this chat interface and you can say, well, I'm preparing for so on and so forth, and it'll make a deck and it'll link it to documents because Myro is trying to become this workspace and it does a lot with the product that I'm kind of responsible for. Where I've seen it be effective is we have a lot of unstructured data. We have a lot of people manually making it into structured data and risk management is all about
four I principles. So we have a lot of people doing that and also double checking that. And that's what we're trying to optimize, right, Going from this unstructured data to something that is structured and then having a person more productive by doing so because otherwise they would have to do everything manually. The bigger the documents, the bigger the manual process, the bigger the gains and the better the business case.
And I've also seen cases where it's like, yeah, we do this thing maybe once or twice a year, and we could automate that. And then people are trying to kind of solve that problem. Yeah, it's like everywhere. And definitely not everything is worth it. But I feel like from the productivity standpoint, it is worth trying kind of everything and seeing what sticks, but not necessarily on the product side when you're thinking of implementing it, yeah. I, I think so.
So that's absolutely true. I think the productivity part of it, right? And we often refer to AIS as Co pilots or agents, right? That's that's the new term because it's always helping you become more productive. And is that, that AI thing is your agent or a copilot. And that is where I think the, the most amount of usage and use cases are coming from product managers as well.
Whereas when it comes to actually building products with AI, the the problems with using data have not gone away the same as it was like a decade back, right? If you had to build a machine learning model on your data or predict the next outcome, let's say your product depends on predicting something about the end customer or end user. Let's say it's a food delivery app, right? And can you predict that this this person would again order
the same thing? Or do they want to order this at this point in time because you have so much data? And I think everything still comes down to do you have meaningful data, enough data, clean data and the data pipelines to basically fetch and, and train your models, whether it's an LLM model now because of generative AI or it's a machine learning model that was predicting whether this user would do again XYZ at this particular point in time.
And I think people jump to using AI because there's just so much available now to, to embed, but the basics of availability of data, the, the meaningfulness of data and the volume of data, I mean, all of that still remains preconditioned. That has not changed. So people, I think that that's, that's where I see a lot of struggle where people want to do something with AI. And then the AI obviously
hallucinates, right? It makes up things because your data is first of all not clean enough to give it all the information it needs. Maybe it is not formatted in the right way or whatever is the issue and then the AI obviously hallucinates or makes up things. Yeah. So I definitely see automation in product. The usual cases like chat bots and stuff are pretty easy to do, I would say, because of the most of the companies are using standard tools, right, for customer service and customer
service interfaces and stuff. But we are not using standard tools for everything else that we do, right. So it's very difficult to get that clean data all the time, yeah. I feel like when it comes to kind of chat bots, but also I was thinking while you were talking, I've read this article about Duolingo and how they've kind of done revenue and the subscription service.
They barely have any subscribers, but they have so many users and still they're very profitable with regards to how many paid subscribers they have versus just general users. And they do a lot with regards to experimenting, what type of notifications get people to start playing and to keep playing and to hook them back. That kind of personalized
approach. I feel like if you have the data to accommodate for that, you can experiment very quickly with regards to AI or how you generate those messages, because right now they have a team of writers that do that as well. And that's something you could generate with regards to personalization. Also in chatbots. If I talk to a chatbot, it's like, yeah. This this is this default entity and it talks the same way to
everyone. And nowadays when a person pops up and I see like a name, I'm like, is this a person or could this be something that's like generated content? Yeah, which is going to be interesting and it's definitely going to be effective, I feel like, because that's what I mean, if you're trying to sell a product, marketing is a thing, right? And you have different marketing strategies. The way you talk to a younger audience is going to be different than the way you hook
in an older audience. And you can experiment, I think fairly quickly with regards to generating whatever marketing content you want to have. Yeah, absolutely. And I think that's why these use cases are sort of catching up
the fastest, right? Because I mean, if the, if the, if the LLM model can understand the tone of different customers, because it read through thousands and 10 thousands of voice, e-mail, SMS, whatever communications across all channels you can understand, OK, this is the tone of this company and this is how we sort of address our customers and stuff like that. So, and, and The thing is the, the AI doesn't bring its own personality.
It sort of builds on the personality that your company or your agents all always had the human agents. So if they have been nicer, the AI could be nicer if they have not been so nice to customers than AI would also because it's the same data. And yeah, it's a reflection, right? And that's where the fine tuning comes in. And the issue is that I think all and obviously you need executive binds because you're investing in these two tools and automation sometimes, right?
Also in, in terms of training your models, it cannot be perfect day one. Like some of the other things that we try to do in product where we at least try to make sure that we validate our hypotheses, we make sure that users want it and then we launch it in, in production and then we say, OK, does it make sense? In case of AI, the iterations are super small. In a regular product feature iteration would be probably once a month or maybe once three
months depending on the product. In case of AI, I've seen the iterations, you can actually literally do it over a period of days or even hours, depending on how soon you get a feedback. And getting used to these faster iteration cycles where you can tweak some parameters of the model, give more weightage to recent conversations, but less weightage to older conversations. Let's see how the conversations go. Ah, this didn't go well because customers did not like a chat board.
Let's change the weightage and parameters of the model. The next day, the model then starts giving another output. So I think that the iterativeness of product, I mean, that's all what we have learned in product. It does. It's so much faster, right, with the AI and LLM models. But you need to be open to that kind of ability in the company and the mindset to be able to do that. Yeah, I also feel like you need
to be able as an organization. Yeah, some organizations are like kind of hesitant in adopting. But if you have a separate organization that is your competitor and they are doing this so you have content generation, then someone picks out a few different snippets and you AB test that and then the next day you're like, oh, let's tweak it because we have high volume to reach some conclusions. Yeah, Yeah. It's like you get outpaced significantly.
Easily, Yeah, yeah. And it all comes down to the mindset and obviously what are you training your models on? I think data, not to forget, right? Data like in the previous boom when there was a software as a service boom, right? And data was the king.
Data is still the king because then that's what your AIS are getting trained on. So I I still feel that the companies who got it right 5/10/15 years back when they were investing into data, investing into the infrastructure of, you know, how the data is getting stored and transferred will still leap forward as compared to companies who did not invest into data and and data infrastructure because they are doing it now. Yeah, yeah.
When all the companies are already sort of doing these iterations on AI models, they're still, they would be obviously a lot of companies, right. And nothing wrong with it. It's just nobody predicted this boom would suddenly come. So a lot of companies were still trying to figure out the data part of it because that that's how your AI would be good.
Yeah, interesting. I feel like looking at what companies are doing now, I feel like all the AI startups are creating tools or more on the tooling aspect of things. And when it comes to personalization, that's where the data aspect really hits. All right. Without data, you cannot personalize. You have no clue what your user base or your customer base looks like. So how can you slice and dice and then personalized content that is generated? Yeah, true. It's not possible.
It's not possible, no. So I think that the AI differentiation is the data differentiation. You can say, right. So you can say that at a fundamental level. And then obviously, if, if let's say the three, three or four companies have good data, then the differentiation is about where can you apply it, how soon, how iterative can you be? Then it's about the
experimentation mindset. And that's where we are seeing companies sort of, you know, moving forward or staying on the back foot depending on where they are in this journey. Yeah, I was thinking from a content perspective actually. I don't know what triggered this, but I saw a demo. A friend of mine knows I do this podcast and he scraped all of the transcripts from YouTube, close to like 190 episodes
nowadays. And he showed me that he could summarize, he could do whatever with regards to the content that I created or what I said and what guests said. I was like, Matt, because YouTube, it's basically, yeah, you can scrape that. It's out in the public. Yeah, yeah. Same for your newsletters. Like what, what do you think of people using that with regards to either honing their knowledge or doing whatever they want with
it? I, I don't have an opinion yet, but I was like, Matt, there's a lot of possibilities I didn't know were possible in the 1st place. Yeah. I truly think I can say about writing more so than than podcasting. I truly think that a lot of writing would not become the differentiation. It will be more about your
personalized experiences. So saying just general stuff about market practices, product and stuff, I can today probably and automate an AI which publishes the newsletter every two weeks for the next one year and probably at a good quality and the tone in which I write. Yeah, it's possible today, I'm 100% sure of it, right. So I can just automate and it starts writing, but the content that's generated is at best average. I mean, that's that's how LLM's
work. So they take all, all of the knowledge and they probably generalize. And do I think it always needs a more average output than amazing output, right. So that's what I've seen. So let's say that table stakes, everyone has the same tools, Yeah, in writing. And maybe same goes could go for
podcasting as well. I don't think generating content, which used to be the big thing, even one year back when we last spoke, I said that the discipline of writing and making sure you do it, you know, consistently is going to be a differentiator. And then the quality is the second step. Yeah, I think the first step in writing at least, will become table stakes. Everyone could generate content at whatever, one every week. You can even do it every day if you have any.
It doesn't matter. You don't need any data. You don't need anything, right? The tool and the tool can with it's average output, it can make sense. Yeah. I mean, you can write on product strategy, product growth, product experimentation and you can generate content every day. So I don't think the the the differentiation when it comes to writing would be any more than the consistency of it, because the consistency has been taken over by these tools and can be taken over.
I think it will come down to the the personal human experiences that we all have in our day-to-day. Yeah, that is so specific to the context you're working in and so specific to the people you work with and so specific to the customer you're serving is what people would want to hear more. So they don't want to hear just only about, OK, how to do experimentation, how to do product strategy.
I think there's just enough that all these tools we talked about can generate and people can self educate. I think it'd be much more about applications and human experiences. And I think the same goes for product careers. And I'm just pivoting here to product careers because that just triggered a thought. I think even in in product management, we talked about productivity. Yeah, let's say after a few months or maybe in a year, it productivity is table stakes.
So everyone is expected to use a tool to become more effective. Just like when Google came, I don't know what was the adoption, but now we all are expected to use Google whenever we want to. It'll be table stakes next year that everyone is just expected to use the tool to become more productive. And then what's your differentiation even in your career, right. Because if everyone has the same Productivity Tools, I think it'll come down to the more impact driving things.
So knowing about your customers, spending time with your customers people skills. Yeah, interpersonal people skills, which AI cannot replicate and everybody cannot have the same kind of people skills. I think this will become the differentiator even for product managers. Yeah. So how do you drive stakeholder alignment? How do you get buy insurance? How do you maintain
relationships in the company? How do you drive impact by, you know, syncing people in a in a in a similar direction, for example, I think these things will become more important than, you know, managing your backlog or is it Scrum or is it Kanban, writing a user story. I think all of these things will become table stakes. Everyone is just expected to be good at them, but they are not differentiators anymore.
No, I love that. Yeah, the the personalized things that you can influence, those will be the differentiating making factors, right. If knowledge becomes a commodity and skills by means of tooling you become productive at them by default, then those are the only things that are going to distinguish me versus another person that has the same kind of skill level or can can learn the quickest I would. Say, yeah, yeah. And the same goes for writing. Exactly.
And maybe we can draw some parallels for podcasting, but I think when knowledge becomes commodities, generating content or generating documents or stuff is then everyone is just expected to be good at that. Then what do you bring in? I think it's it's these more human stuff or soft skills I would say, right that become the differentiators. Interesting when I think it's Google that has this tool, but you can upload a document and it creates a podcast.
People ask me, OK, are you afraid of what you're doing, right? Because look at this, this can do what you do. And then I was like, but my stuff is like a unique experience, right? I get people on and I hear about their experiences. And it's the only reason that I do this is because I enjoy people's stories, right? And their perspectives. And I feel like I doesn't matter where it ends up in my brain. I can't really pinpoint where it ends up, but it helps form my
own perspective. And obviously it's like generated content, right? I can't figure out who said what, but it forms like a basis of a gut feeling that I have. And that leads me to make certain decisions that are others that are different than others. And they're based on the foundation of my conversations with people, right? And their perspectives and their
experiences. I feel like that compounds and that really helps me make decisions even though I don't know where it originates from, but I enjoy that form of learning. And I think that that would become the differentiator, right? And and to be honest, the audience. And you know, in case of broadcasting and writing, we're talking about an audience. In case of if you're communicating it within your company, we're talking about stakeholders and people and partners you work with.
So anyone who's receiving information today, at least what I see and probably it's anecdotal, is that people know when it is an AI generated stuff versus a human generated content. Yeah, you can see it from the quality of the output that's generated, right. So I'm sure the Google tool can generate a podcast versus a real podcast like this where, you know, we are sort of also figuring out where this podcast is going. I think this this is this brings the uncertainty element into it.
And that's more exciting than something which is predictive and you know, can just goes in One Direction. And on LinkedIn, just to give one example, on LinkedIn today, I see that a lot of comments are from just AI bots. You can. See that? Yeah, yeah, it's, I would say, I'm sure LinkedIn will do something about it, but I think we are sort of getting infested by a lot of accounts of people, but also companies who just
commenting on content. And you can get, you can guess that this is a bot because every time you post something on LinkedIn, it's always the same person commenting in a very generic way and in a very formal manner. And you can just see it's it's, it's AI generated.
And then I was talking to my fellow product writers in different parts of the world and they, they confirm that there are AI bots that just commenting on LinkedIn getting traction, getting followership on their pages and stuff like that. And the point I was driving was we can all see, or at least if not today over a period of time, we'll all get it when it's an AI generated content versus a human generated content. The question is which content do people would want to hear?
Will AI get smarter and more interesting and more the surprise element or will the human generated content always has this more, more surprising And because AI is getting trained on the same data that we are generating. So that is yet to be seen but I'm seeing this AI content getting infest infested in our social medias and stuff like that a lot. Interesting.
I I see that. I mean on LinkedIn I didn't even think about it. Maybe it's because I've, I've been out of LinkedIn a lot as of the last few months. On Reddit I see it, but people just call each other out. You're a bot. No, I'm not a bot. And then it just goes like that and I never know what the truth is. So that's why I kind of skipped it in my brain. But on YouTube as well, I see people calling out generated content. But from a surface level, I have no clue, right?
I would have to look into kind of the account specifics and see if it's very generic. But in any case, when it comes to comments and engagements, I feel like that personalized approach and then that uniqueness is what's going to set things apart, right? The podcast that I heard that was that was AI generated, I was like, man, sounds really good, right? Even though it's like kind of formulaic, it resembles things that I've heard in the past.
That's why I was like, man, this, even with the people speaking and kind of the engagements and some of the jokes they made, I was like, it is kind of formulaic. But I have definitely heard shows that are very similar like this, and maybe it's because they also operate under a certain formula that is like more standardized. And I do like these personalized things, right? Either it's an engagement or it's in writing.
I loved your story about Airbnb and kind of the story aspect of how things went and things that happened in the past and then the impact of that. That's what hooks me in. Yeah. And that's, I think it requires a lot of understanding of human sentiment to find such stories to say, OK, this story or this conversation may be of interest to people. Then some generic stuff that you can guess. Where is it going of its direction? Is it going?
Yeah. And that's the next challenge in writing and podcasting and I think even in careers like we talked about, Right. So what can you do that it has this context and individual, individual element of you know, you as a human, as a person, What do you bring in which is so special rather than just this very formula based content or formula based communication, right? I think it will it will become more about that. But this is a bet. Let's see if if AI is getting
trained at a faster pace. But I think for now I can say this so not about consistency but more about quality of content you can generate. Yeah, absolutely. I mean, the, if I look at myself in at a workspace, I can be myself. I can be quite comfortable in being myself. Maybe it's because I'm a consultant and if it doesn't work out, I have, I still have my job basically. So I have like I'm like a wild card and that's why maybe I get comfortable more quickly.
I've been in a session and it was impacting my team and it was impacting others. And then people felt hesitant, right? Asking questions or being themselves in front of a leadership team that might actually impact their career. And I feel like a lot of people can operate like that. More so formulaic, more so objective, maybe not really personalized.
And then it's like, if that's what's going to set you apart, right, this personalized approach, then the workplace, first of all needs to accommodate for that. It needs to be safe enough for people to be themselves, and then the people that operate within that sphere, they need to be able to actually do that as well, right?
And be comfortable in doing so. It's going to be very difficult, I feel like, with kind of generated content and skill levels that are being mediated to a middle just by virtue of being more productive in AI tooling to have that personalization layer on top of it if there's no real good ground for it within an organization. And that's the challenge of, of our times, right for for us working in these companies or even for people who are found finding or starting their own companies.
I think the the first thing to at least do is is to figure out which parts of the, the day-to-day stuff will get, you know, automated or, you know, will get trained on AI. And being in touch with what's happening in AI is just important. So this experimentation mindset, and I think eventually you should find out the repetitive part of your work. Yeah, I was just looking at
something last week. I think it was a conference from Open AI, for example, where it gave a good example about repetitive tasks, Right. So, and, and generally the, the 1st place where a lot of things go is like the customer service part of it, right? That's always where people somehow start. And I, and I now know why because it has the most amount of repetitive things that you could do. There's always an inbound
e-mail. Yeah. And then somebody will look at the e-mail, read it, and then probably either reply or just forward it to someone or just not reply, whatever, Yeah, depending on the context. And you can actually create easily AI agents who can do all of that exactly in the same way. You can just train them that, hey, if e-mail has this, then forward it to this and so on, so on and so forth. So they'll do like a real human conversation with the customer, tell them what do they want.
Sometimes they'll get the job done, and sometimes they'll just escalate it to someone. And these are all AI agents. There's no new, there's no human involved. So if I was in customer service, for example, since we're talking about OK, how will people differentiate in the workplace? I would definitely would want to know what parts of my job are more repetitive and can be, because it's not a matter of
decades. It's a matter of only, I think one or two years that this will mostly impact everyone. I would say in customer service, for example, I would then want to do which, which parts of my job are not repetitive? Yeah, the the part where you bring extreme customers delight, Yeah. I mean, there are a lot of things you can do for customer delight, which is not repetitive because the data doesn't show to AI how can you delight customers and you can delight customers, right.
And then that's just one way to to find out whether it's about helping them out in the, in the ways they did not expect, for example. So that's how I would say that the customer service should think about how to delight customers. And then the customer service is vast because customer service is also not only a way to address customers queries during the usage of the product, but it also tells you what shape the product should take in the future. Because they're hearing a lot
from customers. Maybe they do not like a product feature or they just. I hate some parts of the product. So it also goes into your product road map. So I would also, as a customer service person, I would want to go to the product teams and say,
hey, looked at the data. I used AI to sort of consolidate it. And I think these are the top three features that the product should be looking at because my data tells it. And then you're delighting the product team because you, you're, you're not just throwing, you're not just doing your repetitive task, but you're sort of, you know, bringing more information and insights about customers. So I mean, these are just two examples of, OK, if everyone has access to knowledge, how can you
differentiate? And I think then you then you need to look at it as a product manager, for example, right? We talked about. So I think you just need to go deeper into your job and say, OK, what parts can be taken over by AI? Yeah, that's odd. But this is like AI think a prime example because you have two, I think, user perspectives or at least two people I think of clearly in my mind, the people that when something happens, it's like more so a woe is me and I need to find a
different job, right? Because this thing is going to be automated and people that really like what they're doing and that will figure it out. We'll figure out how they can be that differentiating making factor. Like if a customer calls and you go above and beyond and you help them more than what you're supposed to, and that's the person that's probably going to figure to be productive and to still do that thing because that's what they get excited
about probably likely. And that's going to be the differentiating factor. I was on the phone because I ordered some stuff and it was like taking two weeks. So then I immediately was like, this part can be automated because I call the store, no one's there to pick up. Then I get this other person that cannot find my order and I'm just waiting, waiting. And I called day in and day out and no one picks up. I'm like, for me as a customer, this whole experience is horrible.
And I love the products. I bought the products, but will I do it again? I'm a little bit more hesitant. I might change a few things in the way how I do it. And it all impacts this brand that a company has. So then when you automate things, it's going to be delicate balance, right? Because if automated well, it's going to work well. If automated not well, you're
going to have brand hits. And then indeed the people aspect in the more personalized approach, I feel like that can be a huge differentiating factor. And over here there's an interesting pattern that I see right when we got access to all these AI tools, everyone thought how, how can we apply these AI tools and, you know, apply them to any sort of domain product or, or customer service or
operations. But I'm, I'm seeing this coming up of pre built AI agents, just like how software as a service took over all the software on premise and everything is on cloud and you can just access it on your browser like and you don't need like a software on the CD-ROM or something like that. I think that's what AI will do to the software as a service part of it where you are getting
now pre built AI agents. So let's say, and this is where I'm just sort of taking a bet at the future, right? How it will go is because I already saw some good examples, especially in the Silicon Valley. So I think it will take time until it sort of expands globally. OK, where let's say you are a financial services company and you're looking at hiring somebody who can help with loan servicing. Yeah, mortgage or whatever it is. So loan servicing has a lot of elements, right?
So you have to pick up the call, understand what the customer wants, sometimes restructure the loan, sometimes they want to pay early, sometimes they just want to transfer the loan to something else. Some complex stuff right there they can do. Now you can choose to automate this in your own company, but it takes it's own time. Do you have the data? Is it the right tone?
Will the customer happy? But they're pre built AI agents out there with which you can sort of in, in a matter of weeks can get trained and they can fine tune themselves and people, they're big banks out there who are hiring not real people, but AI agents. And they'll place an order for OK, I need 10 agents because I have 100 customers. Yeah. And I already see a leap from AI as a copilot to AI also has an employee, if you think about it, right, because it's already happening.
They're banks who were sort of hiring these AI agents to take up the customer service stuff. So I think it's going much faster than we expect overall, right? And I think the all of these, and that's where I mentioned the beginning that I thought it was a hype cycle. And obviously there's always hype two things when they come up, right? We are we're optimists and we always hype things more than they they should be. That happened during Web 3 blockchain and always that
happens. But I think this has real world applications. Yeah. And you can see it because people are paying for it and the people who are paying for it are also not betting on it. They're real big companies and real big banks who mean business and they're betting on it and they're investing on it. So I think if I was a product manager, I would already think about not how to use AI to become productive because that is something is already happening today.
I would rather think about what parts of my job are completely replaceable because this concept of AI as an employee or AI getting AI agents to do some parts of your job is just going to happen very soon. So I would just think about, OK, what can I do to get myself trained to do things in to figure out the parts of my job which I can do better than
others. And obviously I'm doing a very, I'm taking a very objective take at this, not a subjective take, so that it makes sense for a larger part of the audience. And then it depends case to case. Not every company is same. Not every company is father in the AI journey. And some companies may just stay as it is for some years because their products don't need AI. Yeah, that's that's also going to happen right there. Exceptions to this? Awesome man, I've really enjoyed
this conversation. I must say I love picking your brain about the product space and especially now that what you've seen in AI as well, I think it's very interesting in how things are going to evolve. And sometimes from a system perspective, I think optimally everything would be integrated, right? You would have one big system because then you can leverage what data you have. But that's not how product works, right?
You solve a specific problem and then you get many products within your landscape as an organization. And I feel like AI is going to be the kind of the the agents that are going to operate these systems and be more productive in that. For sure. And you know, there are a lot of young product managers who reach out via the newsletter or something. I think 1 fundamental that will not go away is just understanding your customers. I think that's the, the something you should not get rid of.
Because the problem with hype cycles is everyone thinks this, this thing is the part of gold and this is where you know, we should all go. But whichever company understands the customers the best, understands their problems the best, knows in what ways we can solve their problems, will still differentiate from companies who are just applying AI for the sake of AI. Absolutely. So the basics don't go away in product management.
Yeah. And what can change is how you collaborate to get the problem solved. But understanding what problems exist, what customers face is still the the sort of the table stakes for any product manager. Yeah, that, that's for sure. Understand your customers 1st and then. No, no replacement for that. Absolutely. Yeah. Thanks so much for coming on. Bond on this was a real joy. Thank you so much. We're going to round it off
here. If you're still here, let us know what you think of the episode, like if you like, dislike, if you didn't like it, and otherwise I'll put all Bond on stuff in the description below also is newsletter. Check it out and we'll see you in the next one.