.png)
Beyond Features: How AI is Redefining the Product Manager’s Role
- JJ Rorie
- Nov 11
- 21 min read
Episode 102
Gopikrishnan Anilkumar, Principal Product Manager at Amazon and member of the Forbes Technology Council, talks about how artificial intelligence is transforming the craft of product management.
Gopikrishnan shares his journey from traditional software product management to leading AI-driven teams at major companies like Amazon, Walmart, and Goldman Sachs. Together, JJ and Gopikrishnan explore what remains constant in great product management—customer empathy, communication, structured thinking, and stakeholder alignment—and what’s changing fast as AI becomes central to how products are built, learned from, and improved.
Listeners will learn how AI product managers differ from traditional ones, not because they code or build models, but because they design experiences that learn and evolve. Gopikrishnan explains how metrics shift (from satisfaction and revenue to latency, model performance, and hallucination rates), how teams expand to include machine learning scientists and evaluators, and why “being AI-aware” is now an essential product skill.
He reminds aspiring PMs that you don’t need to be a machine learning expert to succeed in AI—curiosity, continuous learning, and strong product instincts remain the keys. In the next few years, every product manager will interact with AI in some way—through UX, analytics, content, or internal tools—and those who embrace experimentation and trust-centered design will lead the next wave of great products.
Resources / Connect with Gopikrishnan:
LinkedIn | Personal Website | Forbes Article: AI Isn’t a Feature — It’s a UX Paradigm Shift
Watch:
Listen:
Read:
SUMMARY
Topics Covered:
The evolution from traditional to AI product management
Why PM fundamentals—empathy, communication, alignment—never go out of style
How AI changes metrics, responsibility, and stakeholder complexity
Designing for trust, safety, and explainability in AI experiences
The mindset shift from building features to architecting experiences
The experimental nature of AI product work (build, test, learn, repeat)
Why you don’t need to be a machine learning expert to be an AI PM
How every PM will soon interact with AI in some form
Advice for PMs transitioning into AI and for new PMs entering the field
Key Quotes:
“AI product managers are not building a feature. You are building experiences — you’re an experience architect.”
“Being AI-aware will soon be one of the key strengths of a product manager.”
“You don’t need to be a machine learning expert. You need curiosity, learning agility, and strong PM fundamentals.”
Takeaways:
Trust and responsibility are the new pillars of AI-driven product management.
Product management is evolving from control to influence — guiding systems that learn and adapt.
Continuous experimentation and iteration are essential in AI environments.
AI literacy is becoming as foundational as digital or mobile literacy once was.
TRANSCRIPT
JJ: Hello and welcome to Product Voices. Today we're gonna be talking about. The traditional product manager, if we can call it that, versus the AI product manager and what that means, what that means today, what that we, what, what we think that might mean in the future. Um, is there a real difference? Is there something that we should be debating on or thinking about as product leaders and product teams?
Um, AI is a part of our world and will be, and that goes for both our products and the craft of product management. So as a product manager versus an AI product manager, um, what does that mean and what should we we be thinking about in terms of, of that distinction if there in fact is one? Um, so I'm so excited to have my guest here.
He is a real expert on this and has seen. Seen a lot over his career, so he's gonna gonna add his insights here and it's gonna be a a wonderful conversation.
Gopikrishnana Anilkumar is a principal product manager at Amazon with over a decade of experience leading both AI driven and traditional product teams.
He's worked across companies like Walmart and Goldman Sachs and Amazon, and driving innovation and data platforms and customer facing solutions. He's a Forbes technology council contributor and frequent speaker, and he brings a unique perspective on how AI is reshaping the craft of product management.
Gopi Krishan, thank you so, so much for joining me and sharing your wisdom with us.
Gopikrishnan: Thank you, JJ. Uh, nice to be here.
JJ: You know, I gave a little bit of, of your background, which is a amazing, and, and I love the, the, the companies you've worked for. Um, all very different from each other, but leaders in, in their own right.
And I think that's a wonderful thing. Um, and I, I often find that the stops along our career and the places in which we've worked. Have an impact, right? And, and colors our perspective on product management and especially in AI product management, um, uh, over recent years. So is there a point in your career or an evolution or something in your background that has been helpful in kind of setting your perspective or, or setting your point of view on product and AI product?
Gopikrishnan: Sure, JJ. I think that's, that's a great introduction, uh, uh, to my, uh, background. Um, so just to, uh, give a background about myself, uh, I've been in product management for the past 10 plus years, um, working across both traditional software as well as AI driven systems, uh, in my early part of my career. I worked in environments where business logic was much more clear.
Uh, it was well-defined. Outcomes were predictable. What we wanted to build was also well documented, and it was easy to understand. When I moved into ai in the past, uh, six to seven years have been working as AI product management and in ai in the AI piece, I realized that. One of the key things that I realized that you are not building features.
You are primarily focusing on building systems that learn, evolve, and, and continue to irate and improve over time. So what that means is that it behaves in unexpected ways most of the times. And that has actually taught me about product management specifically. Uh, it has taught me that product management is always not about control.
It's, it's about influence. Uh, and, and that is one of the core themes of being a good product manager also is, uh, you, your job is to guide the system and, and help it learn as much as possible. And, and that has been a, that has been a, that has been so true. It's true in the traditional product management field, but it's more true for.
An AI product management, uh, space. And, and that has been one of the core learnings when I moved from a traditional product management, uh, afield into an AI product management field.
JJ: That's amazing. And it, it makes so much sense, right? To, to think about it in that way and how, you know, we kind of traditional, we're guiding the, the team or influencing the team and, and you kind of put that same, same, um, kinda mindset or skillset to, to the systems as well.
Um, so that, that seems to make a lot of sense to me. So thank you for sharing that. So let's jump into. Maybe the similarities, right? The foundations of product management and, and I'm a, I'm a big believer, um, that, and I've, I've, I've often couched this out outside of the AI conversation, but more, more kind of types of products, right?
Software products versus physical products, for example, and, and how sometimes there are schools of thoughts that those are very different crafts. And the truth is there's a lot of product management that should be the same, regardless. Um, I'm assuming, um, or at least my, my, um, kind of somewhat early hypotheses, um, are that that is still the same, uh, case there.
There's a foundational product management that's going to be the same whether we're quote unquote AI product management or traditional. Um, but I'd love to hear your thoughts, like what are the similarities? What's, what are the elements of product management that remain the same, whether we're talking AI or traditional.
Gopikrishnan: I think there has been four key pillars of product management that I, I can kind of bucket into that stays the same. So the first one, the most important one is customer empathy. Uh, you need to, and that is one of the core, core product skills and, and you need to have a clear empathy towards solving. What the customer problem, you need to understand what customer problems are and, and that having a clear customer, having an empathy towards those problems and solving those problems is always the foundation, um, you need to have.
And that it starts from there. And that's thesis C. Whether it's an AI product manager or a traditional product manager, uh, you need to have that. The second one is you need to be a good communicator. And that is very important and more important, it's very important in the traditional product management, uh, field, but it's more important when you are in an AI product management, especially, uh, written communication, helping align teams, uh, helping able to align teams.
Because you are stakeholders, you, you have different stakeholders which are more internal as well as external stakeholders, and you need to have clear communication. The third thing, uh, is, uh, structured thinking, which I always value in a good product manager, uh, is ability to drive and use data and sometimes, and also use judgment to help drive decision making so that again, stays the same because you need to be able to do that very well, uh, whether you are a traditional product manager or an AI product manager.
And fourth thing, which ties back to the communication, uh, which I can think of, is again, going back to into stakeholder alignment. You need to be able to align across stakeholders and you need to be able to drive that, that that's the, that's the key of building a good product. Having a good alignment among all your stakeholders and, and building, building the right product, uh, will also be, uh, possible only through that alignment.
So I think these four are really important and, and they, those tease to see whether you are an AI product manager or a traditional product manager.
JJ: Yeah, I, I agree with that a hundred percent. I think one of the, the kind of funny things to me is. I wrote a book, it's been out like three years now and three years in product management.
It's like three decades. It's like, it's like dog or cat years, right? Um, it, it feels like forever ago. Um, but the name of it's immutable with the purpose being that the, the five skills I talk about being, uh, you know, table stakes. They, they will not change. We will always need those. And, um, they align very well to, to what you just said.
And so I'm, I'm, I'm happy that. I got a little bit of that right. At least. Um, and I think that the world is changing around us for mostly good, um, in terms of AI and, and technologies that will help us, um, be better. Um, but those things are always going to be so important. And I also love that you said even more important in some cases because I think, I think there's a little hype around how AI is going to.
Almost eliminate, um, in some cases a product manager role or, or some versions of it. And I think that that's, I think it, it in, in kind of, um, it kind of encapsulates the human part of it even more. Um, yeah, moving there. So, um, I, I love that and I'm totally aligned there. I love that, uh, perspective. So. What about the differences?
So when we, when we talk about an AI product manager, maybe, maybe even take us a step back and give, give me your definition of that. What, what do you, when you say an AI product manager, do you just mean a product manager who's. Working on AI products or, or is there something else to it? And then from there, what's, what's the difference?
Like what are some differences between traditional and AI product management?
Gopikrishnan: Yeah. I think, um, the key here, uh, in my opinion is that, um, there are the few differences between the AI product manager and, and traditional product manager. Uh, one of the. Primary differences, I would say is that as a product manager, you always care about metrics and, and you, you want to show that your product is, is doing really well and, and, and being an AI product manager, that those metrics really change.
And, and why, why I say is that like why customer satisfaction, the revenue that it generates is all important, but you need to be tracking more number of metrics. Being an air product manager, you need to consider how well your. Models are performing, or what you are using is performing how, how fast the output is because you are, you have an AI system also behind, uh, your product, which is also operating, and you need to be looking at latency.
And, and the final most important thing is hallucination rate, especially considering generative AI has. Hallucination. Uh, it's very, it's a probabilistic model. It's also important to know that, uh, you need to be also concerned about hallucination. So your metrics change what you, uh, why the traditional metrics are important.
You should also be thinking about these metrics, uh, as well, and, and this, and, and that ties back to a thought process, like what? Is an AI product manager, like what is actually, uh, what is actually so different that your metrics have to have to change? Uh, primarily, uh, in a traditional product, uh, outputs are predictable.
Uh, you can think of like, uh, you have a, so you build a software, you know that how the software is going to operate. Definitely you can test it and ensure that. Uh, the, the software operates as per what your expectation is and you deploy the software. But when it comes to tradit, uh, when it comes to AI product manager, the output becomes more probabilistic in nature.
Uh, it might vary even if the, even if the input is the same that you have tested so many times in your data or your, uh, or your, uh, pre-prod environment when it goes into production. It might have a different output. And, and as models evolve and as, as underlying models become much better, these outputs becomes more, uh, varying also.
So those are the key. That's, that's one of the key foundation why, uh, why a traditional product manager or where the product, traditional product management and, uh, AI product management differ. Differ. And, and that ties back also apart from metrics. One, another thing which is also very important is trust and responsibility.
While Preregister, uh, product manager, you would be thinking about, um, uh, features like you want to launch 10 different features in the next release here. Additional thing that you might need to also think about is how you design your product to the safety and ethics, uh, standards. Also, like how you can kind of plug in responsible AI into your product because you need to be really concerned about that.
So what I would think, uh, in terms of the difference is that while all the traditional product management. Feature, uh, or, or capabilities. And, and what you have is important as a traditional product management. You need to be thinking more as an AI product management. You need to be thinking about what are the new metrics that you should be concerned about?
What is the safety and ethics, uh, that you should be concerned about? So I would say that could be the biggest difference between, uh, traditional product management and air product management.
JJ: Yeah, that, that makes sense. And it's this. Added responsibility and, and stewardship of, of that user experience. Um, and, and you know, what, what that output is. So, um, that's, that makes a lot of sense. What about, um, what about stakeholders and team structure? Obviously stakeholder management, collaboration, very important. We mentioned it being one of the consistent things across all product management. Um. But there may be nuances, um, as, as I can imagine.
Um, tell me a little bit about that and your experience. Like what, what is, what does that team structure look like? What's that collaboration look like?
Gopikrishnan: Definitely, um, there, there is a, there is a change there also. So, uh, you are not just working with as a product manager, you are not just working with your traditional engineering and UX teams.
You should also be working with ML scientists, researchers. Along with evaluators. So giving, giving an example. Uh, in, in one of the products that I had worked, uh, in my, um, uh, with generative ai, we launched the product, or we initially developed the product looking at a lot of quantitative metrics, which the science teams could think about.
And, and, and words were outputting through quantitative metrics, but that actually was not giving the right. Uh, uh, right level of like what, what exactly the customers are expecting. It was not tied closely with the quantitative metrics and, and you, what we had to do was that we understood that it was not just about quantitative metrics, it was also about you need to think about qualitative evaluation.
So human evaluation comes into play. Like sometimes you need to have evaluators who also you should be working with. To evaluate your model or evaluate your product, uh, in a, in a human fashion. And, and you need to define guidelines and metrics so that it kind of ties more closely with your, how your actual customer would be using the product.
So there are, there are more stakeholders apart from engineers ux. You need to be thinking about, as I said, like MS. Scientists, researchers. Designers, human evaluators, and, and then, um, all of these stakeholders also come into, uh, ha have their own point of view, and, and they, those come into play. So I think that is the more, uh, challenging part of being a product management in the AI space is because your stakeholders also increases, uh, while your traditional, uh, stakeholders stay. You have to, uh, you have to be working with all these different stakeholders.
JJ: Yeah, and, and you know, again, one of the things that I hear over and over and. Experience myself in, in product management is, is stakeholder management is difficult. It's, it's, it's our, it's our superpower, right? The collaboration, the diversity of, of people and their perspectives is one, you know, one of the things that makes great products, but it's hard.
We're all humans. We all have different, you know, contexts and. Personalities and way of looking at things and point of views. Um, and, and, and that makes it hard. Um, and so we're just adding, adding layers to that, or, or adding some complexity to that, or stakeholders. But, um, to your point, it's, it's very important because of all of the, these elements.
Um, so, so you mentioned something earlier as you were doing a little bit more of a, an introduction to yourself and your background. Um, I wanna, I wanna ask you a little bit about that, and maybe it's a. Mindset or product thinking or how, however, we wanna coin that, um, about. Not just treating AI as a feature, but you know, a real like system or, or some, something all encompassing.
Maybe, um, not to put words in your mouth, but tell me a little bit about that. Tell me about how a product manager who is going to be in an AI environment, like how they need to. Think about, uh, their product and, and what they're putting out there and, and, and not just think about it as a solution or, or a feature as part of the, the overall product. Tell, tell me a little bit about your thoughts there and how you embrace that.
Gopikrishnan: I think that's very relevant. Um, primarily what I, I believe in is that, uh, ai, uh, especially AI product managers are not building a feature. Um, you are building. More experiences and, and you are more of experienced architects.
Um, and that has, that needs to be the way how we think about your, like you don't build one feature ass. You need to think about an entire experience and how you can architect that experience much better and that, um. That's primarily because, again, going back to our earlier point is outputs are not always the same.
Uh, that means you need to think about any feature you are building or any pro any experience you are building. You need to think about user trust, explainability of that, of that feature, and, and, and you should have a focus around that. So, uh, so how you tie all of this is so important. Into your overall product because your overall product might be functioning as expected.
When you're launching this new feature, you should be thinking about. Use trust for the Laire product. It can't be just for that particular feature because one particular feature user press can break the exp, break the trust for the entire product. So all of this, and it can't be, uh, uh, like that kind of differs from how a traditional product manager thinks about like, okay, I, I'm going to, I'm going to own these sets of features and I'm going to be building my, uh, my, my scope is around these features and I'm going to be building these features.
A, as a AI product management, uh, product manager, you have to be more about thinking about the entire UX paradigms, like how you can think of the entire experience and, and that is important. Um, the second thing which I also, uh, going back to my, the earlier question on the teams, uh, that we work with, uh, or the stakeholders we work with.
Clear thinking, uh, clear thing that we need to be, we need to be very strong is how we can move fast and how we can think, feature or think new, uh, capabilities as experiments. Because ML teams work in experiments. They don't pick up a ticket or a, or a SIM backlog and then start to work on that. Uh, you need to be thinking, you need to have that experimental thinking, how you can brand ab testing and, and then it, it's more like a research.
Some, some things might not work. You may be spending like three months on a particular feature or a capability. It might not work, so you need to. Go back, tune, train, test again, and then come back again. And then maybe it'll work that time. So you, you continue to do experiments, you learn, and that's how you build good, good AI products also.
So those are the two things that I would think, uh, how AI product management and how AI changes product thinking, uh, in, in, in from the traditional product management, uh, thinking.
JJ: Yeah, that's, that's fascinating. And I, I would bet that there's quite a bit of. Culture shock in some cases and, and almost, uh, you know, culture, uh, alignment, um, in some organizations, um, I'm sure organizations like Amazon have, are a little bit more native to experimentation and, and, and, and understanding that, that model.
But there's a lot of organizations out there that do still expect, you know, do a, get b as an output and, you know, and not, you know, don't spend three, three months on something without it having some. Positive output. So I, I'm, I'm hearing a little bit of that from, from my clients and, and my, my network that, you know, it's a little bit of a learning curve and culture shock to, to, you know, truly kind of be in that more research mode while still being in a product mode.
Right? Not, not purely research, but knowing that we do, we're, you know, we, we either win or we learn, right? And, and we don't always get it right, but that's just part of it. So, um, I, I'm sure that's an, an interesting. Part of, of, of folks journeys as well. Um, so I wanna, I wanna ask you, I've got a, a few more questions for you.
I, I love this conversation. It's fascinating. Um, if you were to think about your, your background, um, and you've been, you know, in AI for, for a while now, even while starting in more traditional, um, what is something that. AI surprised you with, maybe it's a positive thing, maybe it was a negative thing. Is there some, some part of your career, some, some instance or moment that like what AI could do or how we approached it or, you know, any kind of AI product management that surprised you one way or the other?
Gopikrishnan: I think the, it goes back to what we talked about. I, it's about how unpredictable these systems can be, uh, and, and how critical. Guardrails, having the right team, having human evaluation, uh, can be, I think that's the, that's the most important, uh, uh, important, uh, learning that I have had, being an AI product manager.
Um, and, and, and, and it go also goes back to being an AI product manager. One another thing that is also important, we talked about experiments, we talked about running experiments and, and, and, and seeing how it, how things work. And we talked about also being an experienced architect. What is also important is that you, you, uh, need to be continuously improving your system.
Um, a, a, a, an AI product, a feature or a traditional product manager, you launch a feature. That feature works. As expected, and you don't have to, unless there are enhancements to the feature or bugs to the feature you go and solve and, and you pick up and you improve that feature. But here, the same product that you might have built maybe may also degrade in its performance.
Over time. So you need to be continuously training. You need to be understanding how or, or understanding your metrics and, and seeing how your product is performing and then iterating over it, irrespective. There is no other enhancement or no features or anything, just the existing product assets. You should be continuously, continuously learning on that.
So I think that is also a very important, um, important learning that I have had. Being an AI product manager, uh, you are always, IM improving and adapting based on the behavior. So that is the, uh, that is the, that is one thing. The second thing, which I also think is also very important is. Um, especially in the current world where new models and new capabilities are coming so fast, uh, you, you need to be, uh, also doing things.
Uh, you need to be much more brittle. You, you can't be a traditional product manager where your product, uh, or your features or your experiences are taking three or four or six months. Timeline, the timeline for launching new features or new exp experiments, uh, um, need to be much more shorter. Uh, you need to be continuously learning and, and improving on it.
So I think that again, goes back to our earlier point of like being, running a lot of experiments and learning, uh, on that. So I think those are the, those are the things that I have learned and have surprised me in the beginning. But now to, uh, it has become part of the product management as well. Uh,
JJ: yeah, it, it kinda gives stake. Or, or lifecycle management, kind of a whole new, whole new spin. Right? Um, it, it's much, much faster and continuous. Um, so, so my final question, kind of two part question I'll start with asking for, for product managers out there today who are listening and their current product managers in what would be considered more traditional, either software or even physical products, um.
What advice would you give them if they're interested in pivoting into AI heavy teams or product management?
Gopikrishnan: I think one mean thing or one, uh, misconception is that you don't need to be a machine learning expert to be a AI product manager. Uh, you need, you need to have curiosity. You need to continuously learn.
Um, and, and as new techniques come up, you need to be open to learning that. Um, one thing which is so important in the generative AI space is that you need to learn how prompting works. So I think that is a key skillset that a product manager should also be doing. Um. And that too. That's, and then, then as I said before, like continue to learn new tools as it comes.
Like what, what, and, and read papers and, and be more conversant with what is happening in the industry assets. And then it combined with your traditional PM skills, you have strong PM skills like communication and prioritization. Combine that and that will, that will help you to move into Air Product Manager first.
JJ: Yeah, I love that. Lean into what you do well and then add, add around around the edges. Um, so same type of question, but a little bit different. Um, of course, I work with a lot of, um, students, grad students, undergrad students, people, people starting their careers, very interested in getting into product management, either from the very beginning of their career or early on in their career.
Um, do you think every. PM will be an AI PM in, you know, 3, 4, 5 years. And really the, the, the second part of that question is, what advice would you give to someone starting their product management learning journey? Would AI be a part of that, regardless of, of where they may end up? What, what are your thoughts there?
Gopikrishnan: Yeah, I think that's a great question. I, I, I believe, uh, every PM will interact with AI in some way. Um. They may not be building models, like every PM may not be involved in building new models or, um, new, um, model capabilities. Uh, but AI is entering ux. AI is being part of analytics. Uh, AI is being part of content generation, internal tooling.
So I believe in the next five years or of, or might be shorter, uh, timeframe. Uh, being AI aware and, and is, is becoming more important and that is going to be one of the key strengths of, of fear. Um. It was similar to how mobile became, like, was so um, popular 10 years back. And, and it was so important for a product manager at that point of time to know about the technology.
So it's very similar to that. So you need to be. Aware about ai, and you need to be using AI and you need to be conversant with AI tools. Uh, but you don't need to be an expert, as I said before, uh, you don't need to be a machine learning expert. You don't need to, uh, you need to understand the impact AI has on your product and how AI can be used as a tool for your product to improve on that.
But, um, so that is how I, I envision how, um, PMs, um, PMs, uh, what, what PMs need to learn, uh, in the next few years.
JJ: Yeah. That's great advice. I appreciate that. I, I think it's important. Um, and, and I love how you mentioned kind of the, the similar to to, to mobile. And even before that, it was like the internet, like, seems like that was, you know, four lifetimes ago.
But I mean, it, it really was a, a, a, a hype cycle. Like, like we're seeing now. Um, I do think there will be more and more. Um, impact, um, some, mostly positive, but definitely some negative. As we, as we as a, a community and as a world and a, a kind of product, product community. Kinda get our grips on this, but, um, I think it's, it's an amazing time to be in product management and I think it's, uh, it's, it's gonna be really exciting to see.
So, Gopikrishan, thank you so, so much for just sharing your journey and, and your expertise. You have. Just a unique perspective on this from all of the expert, uh, expertise and experience in the different places you've been to, to learn. So thank you again so much for, for sharing your insights and, and your advice.
Gopikrishnan: Thanks for inviting me.
JJ: You bet. You bet. Loved the conversation. And thank you all for joining us on Product Voices. Hope you hope to see you on the next episode.