Productside Stories

Women in AI Product Management: Why Representation Shapes Better Products

Featured Guest:

Purnima Bihari | VP of Product at Andgo Systems
11/03/2026

Summary

In this episode of Productside Stories, Purnima Bihari shares her journey into product management and why diversity and representation matter so much in AI and product development. She came to product from engineering and project management after a well-known woman in Canadian tech recognized her potential, and she has made a promise to every mentor since: bring more women in.

Purnima explains how inclusive data and problem selection change AI outcomes. She walks through four places bias enters a product, describes golden data sets and why a missing perspective can slip past an evaluation and then spread through an autonomous agent, and shares a real example from healthcare staffing, where nurse shift preferences only made sense once someone with lived experience explained them. She also covers how her definition of success has expanded, how she hires and builds diverse product teams, and the career advice she gives women in product: treat yourself as the product, find mentors and a sponsor, and do not wait until you are ready. She closes with why this is one of the most exciting times to be a product manager.


Key Takeaways

Representation in AI is crucial for fair and effective outcomes.

  • Purnima cites a research report saying less than 30% of the AI talent pool is women. She argues that when representation is treated only as a hiring metric, it never shows up as a factor in product decisions: who decides what gets built, which problems get solved, what data is collected, and what success looks like.

Data bias can lead to operational inefficiencies and unfair results.

  • A golden data set defines what good looks like for an AI system, so a perspective missing from that data set can slip past evaluation. With an autonomous agent, one missed failure mode influences everything that comes after it, which creates operational inefficiency and unequal outcomes.

Diverse problem selection leads to more inclusive products.

  • Purnima sees bias enter in four places: problem selection, data collection, product design, and evaluation. In her healthcare staffing work, nurse shift preferences only made sense once someone with lived experience supplied the missing context, and that representation is what made the product reliable and efficient.

Mentorship and sponsorship are key for career growth.

  • A mentor challenges your assumptions and helps you build your capabilities. A sponsor advocates for you in the room and helps create opportunities. Purnima’s own start in product came when a well-known woman in Canadian tech recognized her potential before she did.

Product managers should treat themselves as the product they build.

  • Define your value proposition: what problems only you can solve, and what makes the way you see things valuable. Then don’t wait until you’re ready. If you can qualify yourself as a minimum viable product, go for it and iterate. Her three-word shirt exercise, “creative technology disruptor,” is one way to find your words.

Chapters

  • 00:00:00 – Introduction and Career Beginnings
  • 00:00:26 – Meet Purnima Bihari: From Engineering to VP of Product
  • 00:02:01 – The Representation Gap in AI (and Why Hiring Metrics Aren’t Enough)
  • 00:03:45 – Where Bias Shows Up: Problem Selection, Data, Design, and Evaluation
  • 00:06:20 – Golden Data Sets and the Risk of Autonomous AI Agents
  • 00:09:00 – A Real Example: What Nurse Shift Data Revealed About Representation
  • 00:11:00 – Redefining Product Success Beyond Launches and Revenue
  • 00:13:54 – How Purnima Hires and Builds Diverse Product Teams
  • 00:17:15 – Career Advice for Women in Product Management
  • 00:20:00 – The Personal Brand Exercise
  • 00:22:20 – Why Now Is the Most Exciting Time to Be a Product Manager
  • 00:24:50 – Connect with Purnima Bihari

 


Why Listen to This Episode?

In this episode, you’ll get:

  • A clear view of the four places bias enters an AI product: problem selection, data collection, product design, and evaluation
  • A plain-language explanation of golden data sets, and why a missing perspective can slip past an evaluation and then spread through an autonomous agent
  • A real healthcare staffing example of how lived experience changed what the shift data meant
  • A hiring approach for building diverse product teams, including the questions to ask when women are missing from your pipeline
  • Career advice for women in product: treat yourself as the product, find a mentor and a sponsor, and don’t wait until you’re ready

Introduction and Career Beginnings

Nicole Tieche | 00:00:00 – 00:00:25

Welcome to Productside Stories. Today our guest is Purnima Bihari, vice president of product at Anggo Systems. She’s a frequent speaker at product conferences and passionate about women in product management leadership and developing product managers into future product leaders. I’m very excited to have her here with us today to share her experience with our product management community. So welcome, Purnima.

Purnima Bihari | 00:00:25

Hi, Nicole. Thank you for having me.


Meet Purnima Bihari: From Engineering to VP of Product

Nicole Tieche | 00:00:26 – 00:00:40

Absolutely. It’s great to have you. I’m excited for our conversation today. And I love to start out our conversation with how you got into product management. How did it all get started for you?

Purnima Bihari | 00:00:40 – 00:02:01

I’ve been in technology for a really long time, about 16-plus years. Product management was around when I was five years into technology. I actually came from engineering, then moved into project management. Something that had the most profound impact on my career was that my opportunity to get into product management came because a woman who’s very well known in the tech space in Canada recognized my potential and gave me an opportunity. So when someone sees your potential before you even know you can do the thing, it changes the way you think about your career.

Since I joined, I’ve worked with multiple startups and scaleups. I’ve built operating models for three-plus companies, and now I’ve progressed into leadership. And it’s interesting: now every mentor I’ve had has had a promise from me that I’m going to bring in more women. Why? Because when I came in, I realized I was alone, or very few people like me were represented. So how I got into product management is also why I became so conscious of that difference, and why I have such a strong commitment to increasing women’s representation in product, especially in AI.


Gender Representation and Bias in AI

The Representation Gap in AI (and Why Hiring Metrics Aren’t Enough)

Nicole Tieche | 00:02:01 – 00:02:10

Interesting. Tell us about some observations around this lack of representation that you’re talking about within product management, or specifically within AI. What are some observations or things you’ve noted?

Purnima Bihari | 00:02:10 – 00:03:45

There’s actually a research report which says that less than 30% of the AI talent pool is women. And that’s interesting, considering one of the most well-known faces in AI was actually a woman, but we completely skimmed over that fact and went ahead and started talking about Sam Altman. And then the same with the Amodeis, sorry if I’m pronouncing it wrong, the Anthropic founders. I had to look in and find out who the other co-founder was, and I’m like, that is something I had observed. Representation right now is taken as a hiring metric. The thing I’ve noticed and read about and started seeing is that because it was only represented as a hiring metric, it doesn’t show up as a metric we should have while making product decisions.

The observation that was made is that a lot of AI in critical systems, in domains where we do not have that influence, doesn’t represent a lot of people. Clinical trials, for example, have an underrepresentation of women. Now if you look at that as clinical trials, what about AI trials? The problem you start seeing is that there’s not enough diversity in product on who decides what gets built, what problems get solved, what data needs to be collected, what success looks like. You suddenly see that treating representation of women as a hiring metric is completely the wrong way to see it, because it basically means maybe we’ll make decisions which won’t work for everybody. But that’s a simple observation.


Where Bias Shows Up: Problem Selection, Data, Design, and Evaluation

Nicole Tieche | 00:03:45 – 00:03:55

Let’s talk more about how you’re seeing this lack of representation. How have you seen this play out in product management and AI?

Purnima Bihari | 00:03:55 – 00:06:20

If we really look at AI product quality, and product management in general, there are four places where this shows up:

  • Problem selection. If you don’t have women at the table, are we going to select the problems that are maybe relevant only to women, or to other groups? So we have problem selection already, and we form a gap.
  • Data collection. Whose experiences are we representing? I work in a domain where women dominate a certain role, but do we know we have the right amount of data? If you’re just looking at what people click on, it won’t strike you. But a woman’s schedule looks very different from other schedules. A mom’s schedule looks very different from that of somebody who’s recently entered the workforce.
  • Product design. Even in my organizations, when we design for user journeys and exceptions and look at the data, sometimes that bias shows up. People rage-click at different places. How they perceive things is different, because the way you react to a product is what your life really looks like. So it shows up in product design.
  • Evaluation. I’ll give you an example. If the evaluation says a product is 95% complete, they say, “Oh, this is successful.” But it goes out, and suddenly it meets certain women who are in this economic group or this age group and have these stresses, and that didn’t show up in the study. Now you have a system that doesn’t work for them.

Let’s say gen AI, and you’re doing document generation. If it completely misses things that a woman would never say, isn’t that a problem? So it fails validation, and it starts failing your deployment. When do you know something is safe to scale? We talk so much about the safety of these AI systems. Take a simple thing: when you enter any form, you enter your gender. Actually, that’s a bad example to use, but [unclear] still, it always blew my mind that they basically blocked everybody’s account for anybody who had added F or female to the category. You see how simple data, and a presentation of data, with no guardrails, can actually impact the technology we use and how much we trust it. Those would be some very high-level examples.


Golden Data Sets and the Risk of Autonomous AI Agents

Nicole Tieche | 00:06:20 – 00:06:30

Okay. When data isn’t accurately representative, what are the consequences of that? Where does this go next?

Purnima Bihari | 00:06:30 – 00:08:40

When we’re thinking about AI product development, and we’re talking about making the system autonomous, you don’t want to make a system autonomous because it can do the thing. You want to make it autonomous because you can trust what it does. So, for example, in AI product development we’ve started talking a lot about this concept of a golden data set. It’s not just a concept, it’s a technology concept: we create a golden data set. A golden data set is a carefully curated collection of examples and the expected outcomes. This is what good looks like. Then, when an AI system is created, we evaluate whether the system is performing as intended or not, based on that golden data set.

“A golden data set is a carefully curated collection of examples and the expected outcomes. This is what good looks like.”

But here’s the problem. What if your evaluation did not test what you didn’t even think of including? Forget the women part. Let’s say I didn’t factor in a certain birth date, and it didn’t show up, and that goes into my data set. So, for example, let’s say we’re talking about representing women in patient populations, and we have these unusual clinical things which never got caught by an AI system, and then you go into an AI system. With traditional software, you could catch it. You could say, “This is a single workflow. We can change it. It’s a bug.” Somebody would report it. With an autonomous agent, that one action is influencing everything that comes after it.

So let’s say you have an AI agent for staffing, and it doesn’t look at the different constraints. If I was creating a good data set, I would look at all the constraints: the contractual rules, the exceptions. Now, let’s say I didn’t get this data into that golden data set. I don’t even know that it’s not going to work. It’ll just not work once it’s in production. And that means now I’ve created operational inefficiency, and unequal or bad outcomes. That all started because I didn’t think about representation and diversity, so I didn’t catch the failure modes. And if we don’t catch the failure modes, how do you give these AI agent systems autonomy in the first place?

Nicole Tieche | 00:08:40 – 00:09:00

I imagine a data set that doesn’t accurately or effectively represent women would really cause a lot of issues, especially in the healthcare industry that you’re in, where there are so many different healthcare outcomes for men versus women. Have you seen that as a significant issue?


Healthcare Staffing Insights and Leadership Impact

A Real Example: What Nurse Shift Data Revealed About Representation

Purnima Bihari | 00:09:00 – 00:11:00

100%. My current company makes staffing solutions. One of the things that came up during our research is that women, kind of like nurses, preferred a late-night shift. I couldn’t understand why. The data just showed it wasn’t even a gender bias. It just showed people preferred and bid on it. When you look at it, there’s a certain gender bias in it. I can’t find any research report which captures it. Then I was attending another webinar around staffing optimization by another company, and there was a chief nursing officer, and she said, “My daughter, who’s become a nurse now, does the same thing I used to do, which is midnight shifts. They’re the quietest and the calmest shifts that I have in the entire day. I don’t want to deal with the daytime sometimes, because it’s a lot.” In the daytime, they were talking about all the systems that are in play, and you add to it the complexity of [unclear]. You can see what she was saying: “When I walk down the corridor in the morning, I have to answer a vendor call. I have to answer all the patients asking for requests. I have to go back to the system and log in and log out. As soon as I take a midnight shift, all of it disappears, and all I have to do is care for my patients.”

So you can see, when you’re making a solution, that comes out because you had the right representation. I was lucky that one of our design leads, her family is all nurses. She got a lot of data which didn’t exist anywhere. Her mom worked at a different level in a different province. Her sister did. And the thing she would say when she saw it is, “Yep, that’s exactly what they used to do.” And she’d say, “As a kid, I grew up with that.” That’s what I mean. That representation is what makes that AI product reliable, functional, and very efficient.


Redefining Product Success Beyond Launches and Revenue

Nicole Tieche | 00:11:00 – 00:11:10

I can see that. Super important. So how has this impacted you personally, in terms of your mission and your approach to product management now?

Purnima Bihari | 00:11:10 – 00:13:54

When I started my career, like every other product person, I measured my success through launches, the revenue I generated, the size of the teams I built. It still all matters a lot. Obviously, a good product leader is a commercial product leader. They believe that good products deliver measurable business outcomes. And that is still 100% true. But I think my internal definition of what that impact is has expanded. Now I think about success in terms of:

  • What’s the product I’m leaving behind?
  • What are the people I’ve developed? Did it help them as much as they progress later? Did they make the right decisions?
  • Did the decisions I make in an organization outlive my tenure?

And that makes you very intentional.

With diversity, as I said, I don’t see things just as, “Oh, this is a metric,” or “This is a human aspect.” If I really look at AI, I’m looking at what the intentional way of building that diverse thing is. I can get somebody who’s worked in a healthcare system, but I can also have somebody who’s experienced it personally, at a level that matters to them. That diversity matters. Then there’s mentoring people, mentoring diverse people who don’t see themselves in those roles, which has become very important to me. Whenever I get time, I make it a point to reply to every LinkedIn message that comes through my inbox. If there are sales ones, I’ll just send a “no, thank you.” But I’ve had amazing people reach out. They said it looked different. It felt like somebody like them was in a position. They literally came and told me personally, and I’m like, this is what representation looks like.

I’ve also actively started partnering and participating in these conversations: for example, contributing and volunteering your time to create AI frameworks, organizational and operational frameworks, legal frameworks, and even just going and talking about this conversation at the right stage, because somebody’s going to hear you, and that person will get influenced, and that person might be a bigger influence than you were. So if you have a seat at the table, that’s not enough anymore. You have to be willing to use your voice for the right reasons, to help this next generation make the right decisions, so that we can have the right outcomes and the right opportunity. So the next generation isn’t fighting the same battles. It shouldn’t be, “Oh, I still don’t see enough women in product.” No, now I do. It’s just changed how I approach my role as a whole.

“If you have a seat at the table, that’s not enough anymore. You have to be willing to use your voice for the right reasons, to help this next generation make the right decisions.”


Hiring Philosophy and Career Growth Strategies

How Purnima Hires and Builds Diverse Product Teams

Nicole Tieche | 00:13:54 – 00:14:15

I love that. I really hear the passion that you have for developing future leaders, future product leaders, and how they can benefit from your wisdom and carry that on. That’s really cool. And how you take your philosophy and your experience, how does that impact the way you hire and the way you develop teams to ensure representation of all kinds?

Purnima Bihari | 00:14:15 – 00:16:25

Representing women a lot in the product world, I think it’s still important for me to hire the right person. And if I don’t see women, then my question is why? Because I’ve worked with some spectacular women, and you could see there were things that only they could bring to the table. In product, we always talk about customer empathy. My experience working with these women is that the level of empathy they brought to the role was completely different. There was a certain discipline they brought to the role. Product operating models were well executed, well designed. There were complex systems, and a lot more stakeholder alignment. So when I don’t see the right person being there, I ask the questions:

  • Are there enough qualified women? Is that the problem?
  • Do we have the right programs? Do they have access to them?
  • Are they making it to the interview process?
  • Are they getting promoted to leadership?
  • Is it just that we had the right people, but because they’re not at the right level, and that’s your hiring criteria, they’re just not there, because they didn’t get promoted at the same rate as, let’s say, their colleagues?

So there are two problems. One, obviously, is always thinking about what’s the problem you’re solving and who’s the right person for it. And if a woman doesn’t show up, why? What are the different challenges?

The other aspect is that I like believing that product management is an inherent skill. Some people have that curiosity and that discipline that product needs, and they don’t specifically have to come from product management itself. Some really good people I hired were actually from other functions. If product management is such a cross-functional discipline, sometimes you’ll see that gap when you go in. You’ll say, “I don’t know if my team understands, for example, the sales cycle.” Then finding a person who’s not a traditional PM but is really good at understanding how sales works, that’s another thing. So you’re always building a team, filling up those gaps, keeping that diversity and representation on top of it. You don’t want to be prescriptive. You want to hire the right people and set them up for success by giving them the right level of autonomy.

Nicole Tieche | 00:16:25

And diversity in skill as much as other things, right?

Purnima Bihari | 00:16:30 – 00:17:15

For example, this is an industry trend where we are starting to say, “Oh, if you’re not an AIPM, are you really doing it?” Because, as I just said, some PMs are so inherently good. AI is a tool. I would not hire a person because they were an AI PM. For me, it means you know how to use the tool. For me, developing the right person is more important, and I think having just the right attitude. But I’m not dismissing that sometimes an AIPM, even without the domain expertise, is just so good at it that they are the right person for the fit. So understanding exactly what you said, what that representation looks like, what that skill diversity is, what they bring to the table, is always very important when I hire and develop teams.


Career Advice for Women in Product Management

Nicole Tieche | 00:17:15 – 00:17:25

That makes a lot of sense. I understand that. So for our audience listening in, is there specific advice that you have for women who are interested in growing in their product management career?

Purnima Bihari | 00:17:25 – 00:19:35

I have three pieces of advice:

  • Treat yourself as the product. I got that advice from one of my mentors when he was coaching me on how to present myself. He asked me the hard questions. He said, “What’s your value prop?” We spend so much time thinking about product-market fit and product positioning, and how we differentiate from competitors. Maybe it’s a good thing to think about your career like that also. What am I so exceptionally good at? What are the problems that only I can solve, or people like me can solve, and nobody else? And what makes the way I see things so valuable? You don’t need to be good at everything. You need to be good at the thing that you can then explain to somebody coming in. That’s my biggest advice: spend your time finding that truth about yourself.
  • Actively seek mentors, and find a sponsor. Mentors were very important to me going forward. I like mentors who challenge me, who are very straightforward about it and challenge me on my assumptions and what I thought about things, and help me develop my capabilities. But at the same time, if you’re in product management, whether you’re a man or a woman or whatever spectrum you’re on, you need to find a good sponsor. You need to find somebody who sponsors you, talks about you, and advocates for you in the room, and helps you create those opportunities, internally and externally: “Oh, you have to talk to this person.” That’s where your value prop comes in. You have to have something memorable that they can take back.
  • Don’t wait until you’re ready. This is very interesting: product management is hired when there’s a lot of uncertainty and not a lot of information available, but when we approach our careers, we don’t take that same approach. We wait. We say, “Oh, wait until I have this certification.” Very early on, I learned: don’t wait until you’re ready. If you can qualify yourself as a minimum viable product, go for it, and then learn and evolve. Do the same thing: iterate, make yourself better, and figure it out every day.

“Don’t wait until you’re ready. If you can qualify yourself as a minimum viable product, go for it, and then learn and evolve.”

Nicole Tieche | 00:19:35 – 00:20:00

Right. And this goes so much back to the beginning of our interview, how you shared that it was someone else who called out in you valuable attributes and characteristics that would make you really successful in product management. Sometimes it’s hard for us to know what our value is, what it is that we bring. But when there’s a mentor, a sponsor, or someone who can really call that out in you before you might even know it yourself, it’s really super powerful.


The Personal Brand Exercise

Purnima Bihari | 00:20:00 – 00:21:05

Exactly. So my boss at that time made us do an experiment. She made the whole team sit down, and she said, “You need to come up with things which you will put on the back of your shirt and the front of your shirt.” It had to be three words, and those three words had to describe you. What did you care about? I’ve started recommending this. Sometimes it’s good, even if you’re not looking for a job, to put yourself on a resume and ask, “What do I come off as? Who am I as a person?” As soon as you put what have I done, what have I delivered, what am I going to call, then have your brand tags. That’s what you need to evolve, even as you go in product management. But she saw something in me that I didn’t. To date, we’re in touch for the same reason, because I’ve evolved. I’ve changed a lot from when I started. But that’s so important.

Nicole Tieche | 00:21:05

Do you remember what your words were for your T-shirt?

Purnima Bihari | 00:21:10 – 00:22:15

Mine were “creative technology disruptor.” Because I was inherently very creative. I like painting. I do a lot of photography. It was my thing. I used to assist in short films. I went to film-making school. So “creative” was literally the thing, and “disruptor,” because I always wanted to ask, “Okay, I know things are done like this, so how do we do this next? What if it wasn’t done like this, but like that?” That was a constant question. And “technology,” because I came into technology because I wanted to do good things. I wanted to use technology for empowering people, and for me, product management was a way to do that. It took us about half a day to come up with it: what do you know about [unclear], what do you like? And yeah, it was my LinkedIn tagline for a long time, but I don’t know if it makes a good hiring tagline. She made us put it on our business cards. I think I still have a few of them.

Nicole Tieche | 00:22:15

Back when we had business cards, yeah.


The Future of Product Management and Outro

Why Now Is the Most Exciting Time to Be a Product Manager

Nicole Tieche | 00:22:20 – 00:22:30

That’s a really interesting exercise. That’s a great exercise for our product management community to think about. Is there anything else, Purnima, that you would like to share?

Purnima Bihari | 00:22:30 – 00:24:50

One thing I would tell product managers: this is probably one of the most exciting times to be a product manager. It’s so unknown, and everybody’s trying to figure it out, and everything’s a green field. AI is changing things. I know it’s very scary, because there are so many reports. When it started, it was like it’s really hitting the product management profession as a whole. But the way I see it, it’s a really big change. AI is changing how we build software, how we work, what really creates a good product. Some products were so solid that you couldn’t even imagine them ever being taken over, and now you’re suddenly asking, do they even matter? A lot of systems became systems of record, and that was their main business model. But technically now you don’t need them. You need the records; you don’t need the whole bells and whistles that came with them. That’s a simple example. I think that’s why, at a time when we are all being forced to rush into this technology, we cannot form a perspective.

Our responsibility as product managers wasn’t just to adopt technology. It was to really, again, understand what the problems worth solving are, and how technology can help you create that outcome or that value you’re going for. And it’s at a point where we need to all look inside and really think about what this means, and develop our own perspective. The tools are amazing. There are a lot of not-so-good things about them. But if you really look, the tools:

  • Help you experiment really fast
  • Challenge all your assumptions
  • Build something you couldn’t build before
  • Fail, and then there are ways human judgment is required

So I think in this world it’s an exciting time. Go experiment. Figure out what AI is going to fail at. Figure out that uncertainty, but do it on your time. Do it in a way that you actually still enjoy it, and don’t wait for permission anymore. You always wait for permission. There are ways to make things happen. We have an opportunity to shape how products are built, what products are built, what principles and standards will be applied, but also how we work. That’s totally worth an opportunity, and it’s an absolutely amazing time to be a product manager now.

“Our responsibility as product managers wasn’t just to adopt technology. It was to really, again, understand what the problems worth solving are.”


Connect with Purnima Bihari

Nicole Tieche | 00:24:50 – 00:25:05

Absolutely. I agree. That’s great. Well, thank you so much, Purnima, for being with us today and sharing your insight. I do like to ask as well: how can our listeners get a hold of you and connect with you?

Purnima Bihari | 00:25:05

I can share my email or my LinkedIn. I’m very active on LinkedIn.

Nicole Tieche | 00:25:10

Yes. And you said you like to respond to inquiries and connections, so LinkedIn might be perfect.

Purnima Bihari | 00:25:15

Absolutely. And thank you so much for having me. This was a great conversation.

Nicole Tieche | 00:25:20 – 00:25:35

Yes, I appreciate it. I appreciate your time, and thank you to our listeners for tuning in. If you enjoyed today’s conversation, don’t forget to subscribe and stay tuned for more empowering stories from the world of product management. Until next time, I’m Nicole Tieche, and this is Productside Stories.