Skip to content

By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

Aakash GuptaAakash Gupta

How Hims & Hers Uses Evals to Let AI Talk to Patients (CPO Interview)

Dheerja Kaur, the first CPO of Hims & Hers, shows how her team put an AI agent at the front door of healthcare, with doctors in the same chat and an LLM judge checking every answer. Then she covers how she hires AI PMs.

Aakash GuptahostDheerja Kaurguest
Nov 17, 20261h 3mWatch on YouTube

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

Great conversations run long. Their best ideas don't have to.

uListen is free on iPhone, iPad, Android, Chrome, and Claude.

See how

Transcript

0:00 – 1:26

Intro

Aakash Gupta 0:00 So the topic everybody keeps talking about is evals. It's pretty much the new PM skill that everybody needs to master. How do evals work in this context?

Dheerja Kaur 0:10 You know, we go through kind of like the classic loop of eval-driven development, where you're testing, you're releasing. We also have an LLM judge that's live in production that's scoring an answer before it even gets sent to a customer.

Aakash Gupta 0:20 So you still have a role that is AI PM. Why is that a different role? Why isn't every PM just an AI PM now?

Dheerja Kaur 0:26 You have to understand the technology at a deeper level and how to work with, with those engineers, and it just looks very different than, like, historically product development looked like. So that's the difference really is, like, are you actually building an AI product, or are you a PM who uses AI?

Aakash Gupta 0:39 How could I become a successful AI PM in a regulated industry? Before we get into today's show, please take a second to check that you're subscribed on YouTube and following on Apple and Spotify Podcasts. If you want access to all of my favorite AI tools, I've gotten them to give you an entire year of their paid plans. Check out bundle.aakashg.com for an entire year of Bolt.new, Airtable, Speechify, Descript, Magic Patterns, Linear, Dovetail, Arise, and Mobbin. And now, into today's show. [fire crackling]

1:26 – 5:07

Why regulated industries are taking to AI first

Aakash Gupta 1:27 Dheerja Kaur is the CPO at Hims and Hers. Before that, she was CPO at The Skimm. She was the first engineer PM at Grantland, and she's one of the few consumer leaders that has shipped in finance, media, and healthcare. Dheerja, welcome to the podcast.

Dheerja Kaur 1:45 Thank you. I'm excited to be here.

Aakash Gupta 1:47 So those three categories that you've shipped in, they seem to be adopting AI pretty quickly.

Dheerja Kaur 1:53 Yeah.

Aakash Gupta 1:53 What's your take on that-

Dheerja Kaur 1:55 Yeah, for sure

Aakash Gupta 1:55 ... AI adoption in these regulated industries?

Dheerja Kaur 1:58 Yeah. So first I'd just like to say my joke about my career is that I've worked on the vegetables-

Aakash Gupta 2:02 [laughs]

Dheerja Kaur 2:02 ... 'cause these are things that people don't naturally like to engage with.

Aakash Gupta 2:05 Yeah.

Dheerja Kaur 2:05 Like, you'd probably rather be on, like, TikTok than engage in your finance app or your health app, right? So it's a hard consumer problem, right?

Aakash Gupta 2:12 Yes.

Dheerja Kaur 2:12 It's like, how do you build delightful, engaging consumer experience in a category that people are inherently have, like, a negative relationship with sometimes?

Aakash Gupta 2:20 Yeah.

Dheerja Kaur 2:20 Um, I don't have, like, the perfect secret sauce, but it's been, like, a, a kind of common theme across my career. In terms of the AI piece, so I do think that finance and healthcare have a lot of overlap because if you think about it, one is these are systems that are super archaic traditionally, very inaccessible. Like, you think about the markets. You know, when Robinhood first launched, it was the first time that it didn't cost $10 to place a trade. Every single trade used to cost $10 before Robinhood.

Aakash Gupta 2:47 Yeah.

Dheerja Kaur 2:47 Um, commission-free trading, massive democratization, right? I probably don't need to tell you, like, how bad the healthcare system is and, like, how archaic that is and how hard it is to navigate. And so I think what's happened with AI and LLMs, like no surprise to everybody, is that this technology has suddenly enabled you to break down these massive barriers and enable what really rich people get in these industries, like financial advisors or concierge doctors at scale, right? The quality of that experience at scale in a way that's super accessible. And so I think everybody [chuckles] has their own experience with AI in health especially, but I think health is, like, the number-one topic in OpenAI, Anthropic. That, that's-

Aakash Gupta 3:27 Yeah

Dheerja Kaur 3:27 ... it's the number-one use case-

Aakash Gupta 3:29 Mm-hmm

Dheerja Kaur 3:29 ... for, for AI because it is so broken, and suddenly you finally have a thing, a technology that can actually dot connect across a very archaic and broken system.

Aakash Gupta 3:40 Why is it, though, that it's almost showing up in these regulated places before the fun stuff?

Dheerja Kaur 3:46 Yeah. I mean, I will say I think from that perspective, finance and healthcare are a little bit different 'cause finance I think, I don't think we've yet seen, like, the killer AI product in finance. Like, I think everybody's dabbling it, in it. But in health, I think it's just this, like, acute problem that everybody feels every single day, right? And so suddenly you have this product where, I mean, I'm sure we're all doing this, like, you can talk to it about, you know, how you're feeling every single day, your symptoms, the different components of what you're experiencing across your health, and it is truly and sadly the only thing you can interact with that can actually help you across that journey, right? Because our healthcare system is extremely siloed. You have doctors and specialties who don't think beyond their specialty, right?

Aakash Gupta 4:29 Yeah.

Dheerja Kaur 4:29 You have systems where you have to go to different offices, different, you know, urgent care or, you know, specialists, and there isn't a single thing, a single, like, entity, a single body that is looking at you holistically, and finally you have that. So I just think it's the acuteness of the problem, right? Like, it just is, right? I don't... I think that that is the number-one problem in America today.

Aakash Gupta 4:49 The same reason it's regulated is the same reason people are turning to AI.

Dheerja Kaur 4:53 Yeah. I think it's regulation. I also think in healthcare it's the incentive system. You've got insurance. You've got the entire system crafted around incentivizing, like, efficiency and siloing, you know? And again, LLMs can completely break

5:07 – 9:06

Why Hims & Hers is going AI-native

Dheerja Kaur 5:07 that model.

Aakash Gupta 5:07 So Hims and Hers has undergone a pretty big transition-

Dheerja Kaur 5:11 Yeah

Aakash Gupta 5:11 ... from kind of point solutions for very specific healthcare needs, maybe I wanna solve this particular problem that I have, to more of an AI-native-

Dheerja Kaur 5:22 Yes

Aakash Gupta 5:22 ... full solution. Why that shift? Why you guys? What's the mode around that? Why have you taken this direction?

Dheerja Kaur 5:31 Yeah, for sure. Okay. So when I first, you know, talked to Hims about joining, I didn't fully understand what the company was going after, and I had this magic moment that made it so obvious to me that if anybody is going to actually solve this problem in healthcare, it's at this company. And that, that's not just a pitch. Like, I really do believe that. So the way I would think about it is that we believe there are three components to solving the proactive longitudinal- Healthcare problem for your average consumer. The first is diagnostics, right? Like, we need to get a picture of your full body, right? And that's across blood testing and biomarkers. It's across wearables data. It's across, like, your medical history, even more advanced testing, like genetic sequencing, right? There should be this, like, living, breathing picture of your health, and unfortunately, it's really hard to get that today, right? I'm sure, you know, pre-AI, I think everybody had, like, Google Drive folders [chuckles] with, like, you know, their different documentation and fragmented data everywhere, your wearables, your, you know, history in different places. And so, you know, the-- being able to create, like, a single living, breathing home for all your health data is really powerful. And for us, our big investment has actually been in verticalizing lab testing. So we are able to ship you an at home blood testing device, send it back to our own lab processing facilities, and own the full data loop there so that we can actually get this, like, ongoing picture of your health. So that's the first. The second piece is the actual medications. Like, how do we actually improve your health? And this is actually where the company has put the most investment historically. So we are fully verticalized in that as well, so we offer treatments across all these different conditions, like mental health, weight loss, hair loss, sexual health, et cetera, hormonal therapy. And we, uh, do-- uh, have the ability for us to actually personalize those treatments. So we can say, "Hey, based on your l-blood test, here's the exact dosage of cholesterol medication that we're gonna prescribe for you. And then as your levels improve over time, we can actually change that dosage over time too." Or we can combine what you need, cholesterol, supplements, you know, uh, weight loss drugs. Like, these are all things that we're able to provide, like, a full stack solution for and be extremely personalized. I think of it as almost like precision medicine, you know? And so that full loop is incredibly hard to build and find, right? [chuckles] Like, having the, the health data, including the, the full stack and the verticalization of, like, blood testing, in addition to the verticalization of treatments and, and being able to actually offer the thing that helps people improve takes years and years of capital investment to build that capability. Hims has that today. And then the third piece is what you're getting at, which is the actual, like, care doctor experience, right? And we have, you know, over 1,000 providers on our platform today who prescribe and diagnose, uh, for these different treatments and medications. And we wanna build-- we're building the AI layer on top of that, which is basically like, you know, "Hey, can we take the best of both worlds?" Where you have what LLMs can do in terms of doc connecting, providing proactive support, collect-- looking at that closed loop of data and saying, "Hey, what does that mean for what your treatment should be?" And then you have the real humans in the loop for the actual diagnosis and prescribing and that kind of, like, human support. We ha-- Again, we have all of those components. And so the consumer experience that we're building is, you know, imagine a world where you're a Hims or Hers member, and on an ongoing basis, you have ongoing, like, lab testing and diagnostics. You have your hyper-personalized treatment plan, and then you have this what feels like a concierge doctor in your pocket with AI and real doctors in the loop. And again, I think we're one of the only companies who actually has the capabilities to build all of that, and a lot of that we've already built today, which is super exciting.

Aakash Gupta 9:05 AI is

9:06 – 11:12

Ads

Aakash Gupta 9:06 making it faster than ever to build and ship products. But deciding what to build? That's becoming the real bottleneck. And part of the problem is that everything that teams need to make that decision is disconnected. Customer feedback lives in one place. Prioritization happens somewhere else. The roadmap gets rebuilt for leadership. And the work engineering is shipping lives in Jira. So when priorities shift, the roadmap is already out of date. And when someone asks why an idea made the cut, the evidence behind that decision is buried somewhere else. Jira Product Discovery brings those pieces together, helping teams make better informed decisions, build living roadmaps, and connect their plans directly to delivery in Jira. Let me show you what that looks like. This is my Discovery project. Every idea has the insights behind it attached, things like customer quotes, support tickets, and usage data, along with the fields my team uses to prioritize. Here's the roadmap. Same ideas laid out on a quarterly timeline. I didn't rebuild anything for this. It's a view on the data I'm already keeping current. To share it with leadership, I publish it. They get a link to the read-only version. They don't need a license to open it. And they only see the fields which I choose to show. And because the link is live, when priorities shift or something moves next month, the roadmap they already have updates too, so no more roadmap underscore V7 underscore final. So that's the living roadmap. The second piece is connecting Discovery to delivery. Say this idea is committed for next quarter. I open it, go to the Discovery tab, and create the delivery ticket right from there. It lands in an epic in the engineering team's Jira project, already linked back to the idea it came from. As engineering moves that work through their sprints, this progress bar fills in. Add the delivery progress field to the roadmap view, and the link leadership already has shows where each item actually stands, so nobody has to ping me or set up a status meeting to find out. That closes the loop. The reasoning stays with the idea, the roadmap is a live view of it, and the work engineering ships stays tied to why it was built in the first place. If your team already runs delivery in Jira, try Jira Product Discovery. Link in the show notes.

11:12 – 14:45

Inside the Hers app: AI and doctors in one chat

Aakash Gupta 11:12 Walk us through it.

Dheerja Kaur 11:13 Yeah.

Aakash Gupta 11:13 What does the provider experience look like? What does the patient experience look like? How does that come together in the app?

Dheerja Kaur 11:18 Yeah. So one piece to contextualize is we're very early in that new consumer experience journey. So a few, a couple months ago, we launched Greenfield new app, and I can talk about some of that too, which is super fun. Like, how often do you get to build, like, a Greenfield zero to one app at scale-

Aakash Gupta 11:35 Mm-hmm

Dheerja Kaur 11:35 ... you know? Um, we launched a whole new app, and we started with hers and weight loss since that's actually the most complicated journey that we see today. Side effects, like plateauing. It's like, it's, it's, it's a tough journey that people are on, and it's just, like, the very beginning, right? So we started with this, like, single use case, and then obviously we're expanding from there into that, that full picture that I described. And so the experience that we built was basically, hey, super simple consumer product. Like, this is a two-tab app, [chuckles] you know? Your home screen is what we call a group chat with your agent and your doctors, where it feels like a single-threaded conversation. And it is a single-threaded conversation where- You're getting proactive support from the agent, you're getting responses to your questions, and then any time it requires, like, a real doctor's guidance, they just are part of that conversation. There's no switching, right? Where I think today sometimes people are in ChatGPT or Claude, and then they have to, like, copy-paste that and bring it to their doctor. This is just, like, one seamless consumer experience. Um, and it's pretty magical. And then the other part of it is, like I mentioned, this sort of, like, living, breathing picture of your health. And so for Hers weight loss customers, we actually ship you a free Bluetooth scale as part of your membership, and that auto-syncs to your AI companion in your app.

Aakash Gupta 12:50 Mm.

Dheerja Kaur 12:50 And so, you know, you step on the scale, you know, if it notices that, "Hey, you're, you might be plateauing," it reaches out to you. "Hey, maybe we should consider talking to your doctor and changing your dose. Let's bring your doctor into the conversation." Or, "Hey, how are you feeling? Like, how are those side effects doing?" You know? It's just a very, like, kind of high-touch, proactive experience that's connecting your data, this agent that feels like that companion, and again, doctor's right there in the conversation, so you're not, like, bouncing between all these different places. And again, it's very magical. I [chuckles] am using that word a lot because, like, I, I sit there every single day and I read all the conversations that, uh, people are having, and that... I can't express enough how amazing it is to have agents and doctors, like, in one conversation. No, you know, passing between, no context switching for you as a customer, and it's pretty awesome. So that's the consumer experience. And then on the provider side, so for those of you not familiar with healthcare, there's a term called EMR, which is kind of like an industry term for essentially, like, how doctors look at a patient chart, right? When you go to the doctor, you probably see they have, like, your chart up, and they're looking at your history and all of that. And so we have that tool for our doctors. It's a big part of what we build.

Aakash Gupta 13:57 So electronic medical record?

Dheerja Kaur 13:59 Exactly, yes.

Aakash Gupta 14:00 Mm.

Dheerja Kaur 14:00 Um, and so we've always had that, but now what's happening is that we've integrated AI into that product for doctors, so there's a super streamlined summary of the patient's medical history, their current treatment plan, you know, anything to know about side effects that they've experienced or things like that. And then it also tees up what's happened in the conversation so far. So in that exact moment, the doctor's able to say, "Okay, got it. Understood. Let me respond in a way that makes it so- clear that I actually have the right context."

Aakash Gupta 14:27 Mm-hmm.

Dheerja Kaur 14:28 Um, and then the best part is, you know, when it, when that goes into the conversation, and again, you'll hopefully see this in the demo, the agent actually then, like, takes the handoff back from there. It's like, "Okay, do you have any follow-up questions?" Things like that. It's really cool, and it's just, uh... I'm, I'm really proud of it, and again, like I said, it's just the very beginning.

Aakash Gupta 14:43 Hmm.

14:45 – 20:54

Protocols as code and a live LLM judge

Aakash Gupta 14:45 So obviously there was a team of PM, engineer, designer, and I guess this other key team member-

Dheerja Kaur 14:51 Yes

Aakash Gupta 14:52 ... doctor-

Dheerja Kaur 14:52 Doctors, yes

Aakash Gupta 14:53 ... building this agentic product, which I imagine you had a lot of, or I would as a product leader, have a lot of nervousness around making sure that it's built correctly. So walk us through how that collaboration works and how you actually build this type of a product.

Dheerja Kaur 15:08 Yeah, absolutely. So many layers to this. So I think first it's important to understand that, you know, this company's been around for years and has, like, 10 years of clinical data and outcomes to train off of. That's massive, right? 'Cause we're at scale. We're serving millions of customers. And so, you know, you're not starting from scratch, which is massive [chuckles] in the LLM space, obviously. And then big part of the way we build products in general, but especially this product, is that we have a clinical team. So we have, uh, medical leads in each of our specialties, weight loss, hormonal therapy, et cetera, um, and they're just, like, experts in their field. They're incredible. It's actually really fun for PMs and engineers, right? Because you get to sit side by side with them, and you take what we call, or what they call, clinical protocols, right? Which is, like, documented guidelines for each different category, the treatments. You get really detailed on this stuff, right? Like, think about things that are, you know, d- a doctor's training codified into writing, right? And then you get to, like, then turn those into essentially code, right? And so, you know, uh, Mo, Mo likes to say this a lot, he's our CTO, that, you know, essentially we treat clinical protocols as code. They get checked in. If there are ever changes to make, you know, that becomes [chuckles] checked in as well.

Aakash Gupta 16:20 Mm-hmm.

Dheerja Kaur 16:21 Um, and so, like, even the d- the, our clinical team gets sort of trained on, like, how to write, how to kind of be an engineer. And obviously a lot of the PM and engineer's job is to, like, actually translate that, right, in the right ways. But they're sitting side by side, like, every day, every week as they're building. And then once this thing's, you know, is getting into dogfooding, you know, there's, like, a daily clinical standup to triage what's going on, what we're seeing. There's a whole set of protocols around that, and they're truly, like, embedded on the team and part of the cross-functional team.

Aakash Gupta 16:48 So the topic everybody keeps talking about is evals.

Dheerja Kaur 16:51 Yes.

Aakash Gupta 16:52 It's pretty much the new PM skill that everybody needs to master. How do evals work in this context?

Dheerja Kaur 16:59 Yeah. So we, I mean, very common in this space, like, you know, we follow eval-driven development, right? And I think, you know, that's a definitely a key part of the PM job. We also have an entire evals team on the Applied AI team as part of that.

Aakash Gupta 17:11 Mm.

Dheerja Kaur 17:11 We, you know, we go through kind of like the, the classic loop of, of eval-driven development, where you're, uh, you know, you're, you're testing, you're releasing. We also have an LLM judge that's in, that's, that's live in production that's scoring an answer before it even gets sent to a customer. And so we have a couple different components of that. I would say it's also more than evals, to be honest. [chuckles]

Aakash Gupta 17:30 Mm-hmm.

Dheerja Kaur 17:31 So, um, one of my favorite examples for our AI PM team is that when we first started, you know, playing around internally with the V, you know, zero of this AI companion, we all felt like it was just really cold and very clinical 'cause it was basically built off of just these clinical protocols. We hadn't yet kind of built the, the actual, like, human layer of that and the consumer experience that we wanted to create. And so one of our PMs, just as one example of many, right, put together, like, a personalization rubric, you know, for how do we actually define what feels like the type of personal conversation we wanna have with our customers. And so that's anything from, you know... And that, and that's across, like, nine different dimensions. [chuckles] So anything from does it know you? Is it aware of your context? Does it know your treatment plan? Does it know your journey? Does it know your goals, the emotional context that you're on? The second is kind of mapping that to the clinical protocols and the guidelines and synthesizing the information in the right way. And then the third is, like, voice and tone. What type of warmth are we looking for? And that actually varies between our two brands even. And so those are the types of things that I see our PMs also doing, not just evals.

Aakash Gupta 18:34 Mm.

Dheerja Kaur 18:35 You know? Um, it's such a key part, uh, to me of how do you become a consumer PM that's AI. Right

Aakash Gupta 18:40 The two brands being Hims-

Dheerja Kaur 18:42 Hims & Hers, yeah

Aakash Gupta 18:42 ... and Hers

Dheerja Kaur 18:43 Exactly. Or even in-

Aakash Gupta 18:43 Two different experiences

Dheerja Kaur 18:44 Yeah, and even between the different treatment journeys might, you might have different approaches, different tones. I mean, that can get a little bit complicated, and there are trade-offs to that. But, you know, the reality is somebody who's on a weight loss journey might look different from somebody who's on, like, a peptides product-

Aakash Gupta 18:58 Mm-hmm

Dheerja Kaur 18:58 ... who's different from somebody who's on, like, hormonal therapy. So how do you work that in to... and codify that?

Aakash Gupta 19:03 So it's a lot more than evals, but let's go-

Dheerja Kaur 19:05 Yeah

Aakash Gupta 19:05 ... deep on evals-

Dheerja Kaur 19:06 Yeah

Aakash Gupta 19:06 ... for a second. Where do they start? Are doctors writing the first evals?

Dheerja Kaur 19:10 Doctors are not writing evals, right? So doctors are basically codifying the clinical protocols, like I mentioned.

Aakash Gupta 19:15 Mm-hmm.

Dheerja Kaur 19:15 And those are essentially checked in. And then between our PM team and our evals team, those are turned into evals, essentially.

Aakash Gupta 19:21 Okay.

Dheerja Kaur 19:22 Yeah.

Aakash Gupta 19:22 So what does the evals team focus on, and how do the PMs collaborate with them?

Dheerja Kaur 19:27 Yeah.

Aakash Gupta 19:28 Let's make it more concrete. So like 100 responses, like these are good responses, the tone here is good. Like how-

Dheerja Kaur 19:33 Exactly, yeah.

Aakash Gupta 19:34 Okay.

Dheerja Kaur 19:34 So, so basically you're writing your evals based off of, like, the clinical protocols, for example, right?

Aakash Gupta 19:38 Mm-hmm.

Dheerja Kaur 19:39 Like, how do I define what is accurate from a clinical perspective, and then you're layering around these additional dimensions, like this personalization rubric. So we have a rubric. Let's actually create an eval off of that, essentially.

Aakash Gupta 19:49 Mm-hmm.

Dheerja Kaur 19:49 Does that make sense?

Aakash Gupta 19:50 Mm-hmm.

Dheerja Kaur 19:50 And then you're essentially in real time, right, as you're s- as you're running these things in the wild, you're essentially, like, you know, live evaluating it and, and then iterate on your evals based off of that.

Aakash Gupta 20:00 Mm-hmm. So your offline evals are really important when you're, like, developing-

Dheerja Kaur 20:04 Yeah

Aakash Gupta 20:04 ... the first version of the product, but after that, you're mainly focused on what's live-

Dheerja Kaur 20:09 Exactly

Aakash Gupta 20:09 ... how are people experiencing it.

Dheerja Kaur 20:11 Yeah.

Aakash Gupta 20:11 Okay, got it. And w- does the evals team, are they still managing that once it goes live?

Dheerja Kaur 20:17 Yeah.

Aakash Gupta 20:17 What is the PM's sort of collaboration model with them?

Dheerja Kaur 20:22 I mean, it's a small team. [chuckles]

Aakash Gupta 20:23 Yeah.

Dheerja Kaur 20:23 Right? So we, when we say evals team, like we have one applied AI team-

Aakash Gupta 20:27 Okay

Dheerja Kaur 20:27 ... across, like, PM, engineering, and that, you know, there's, you know, there's an evals component of that team. You know, there's the, you know, et cetera. So-

Aakash Gupta 20:35 Mm-hmm

Dheerja Kaur 20:35 ... I mean, again, we're... I, I would say this feels very much like a startup, right?

Aakash Gupta 20:39 Yeah.

Dheerja Kaur 20:39 And so we have, like, a small, a small group of people, daily standup, looking at conversations, but also looking at, you know, LLM judges, looking at the evals, looking at what's being triggered in production, and, like, triaging them every single day. And then we have a lot of internal dogfooding as well.

Aakash Gupta 20:53 Hmm.

20:54 – 21:42

Results: 80% resolved by AI, 3x engagement

Dheerja Kaur 20:54 Hmm.

Aakash Gupta 20:55 What can you share of the early success of this product?

Dheerja Kaur 20:57 Yeah. So it's real exciting. Like I said, you know, we're only a couple months in, rolled out as an experiment, right? Not at fully at scale yet. But we're already seeing 3X the customer engagement-

Aakash Gupta 21:06 Mm-hmm

Dheerja Kaur 21:06 ... which is, like, kind of insane. [chuckles] You know? And we're seeing the AI companion resolve 80% of conversations without needing a human in the loop, which is just massive. Like, to have that on day one is kind of crazy.

Aakash Gupta 21:18 Mm-hmm.

Dheerja Kaur 21:19 And that's even before we've really started to scratch the surface around being proactive, not just reactive, right?

Aakash Gupta 21:24 Yeah.

Dheerja Kaur 21:25 'Cause what we've started to do since launch is that the companion is now proactively reaching out to you, like I said, if it notices that your weight is plateauing, or, "Hey, you hit a celebration," or, "You hit a milestone. Let's celebrate that."

Aakash Gupta 21:35 Hmm.

Dheerja Kaur 21:35 "Let's check in on how you're doing." And that's really where your engagement loop comes into play, and we haven't even scratched the surface on that, so it's super

21:42 – 28:12

Specs in Git and Wild West mode

Dheerja Kaur 21:42 exciting.

Aakash Gupta 21:43 So talk to me about the team behind this, how they work.

Dheerja Kaur 21:46 Yeah.

Aakash Gupta 21:46 How has that changed? Are you guys embracing sort of the AI-pilled way to work?

Dheerja Kaur 21:52 [laughs]

Aakash Gupta 21:53 And what are the components that have been most successful for you?

Dheerja Kaur 21:55 Yeah, for sure. So I would say this team especially, I mean, these are all AI folks at their core, right? Across engineering and PM. Doctors are now getting AI-pilled too, you know? Um-

Aakash Gupta 22:06 Oh, okay

Dheerja Kaur 22:06 ... they're, right? 'Cause they're, they're all, they're all operating in kind of the new-ish way of working. I say new-ish 'cause, you know, I feel like we're all there. But yeah, I mean, it's kind of some of the stuff I think you've talked about in your podcast, right? So if you talk to the team, you know, they're not using, you know, classic project management tools or classic, like, sprint planning processes as much as they are operating extremely lean, extremely fast. They have, like, a Git repo where PRDs, if you even wanna call that anymore [chuckles] you know, specs are basically committed into Git. Sometimes you're just auto-generating PRs ba- based off of that, right? And you're-

Aakash Gupta 22:37 Whoa

Dheerja Kaur 22:37 ... you're operating in, like, a super, super kind of, uh, uh, code-based environment, you know?

Aakash Gupta 22:42 Mm-hmm.

Dheerja Kaur 22:42 Where you're getting from, like, spec to code in hours, if not minutes sometimes.

Aakash Gupta 22:46 Hmm.

Dheerja Kaur 22:46 You know? And obviously you have all the checks and balances on top of that. So it's fun.

Aakash Gupta 22:50 Specs in GitHub, huh?

Dheerja Kaur 22:52 Yeah. Oh, yeah.

Aakash Gupta 22:52 What does that look like?

Dheerja Kaur 22:53 Actually, most of our teams are doing that now.

Aakash Gupta 22:55 That's crazy.

Dheerja Kaur 22:55 Yeah, yeah.

Aakash Gupta 22:56 So is planning run out of GitHub too?

Dheerja Kaur 22:59 Planning has not been solved yet. Let's put it that way. [chuckles]

Aakash Gupta 23:01 Okay.

Dheerja Kaur 23:01 I think the, like, you know, quarterly, yearly planning process around business goals, you know, what our roadmap bets are and all of that, that's still a little more traditional.

Aakash Gupta 23:10 Mm-hmm.

Dheerja Kaur 23:10 I think what we're seeing is, like, on-the-ground product execution is getting a lot more similar to, like, what you talked about in your past podcast, like a Team OS type, type flavor.

Aakash Gupta 23:19 Yeah.

Dheerja Kaur 23:19 Yeah.

Aakash Gupta 23:19 What is, what have been the benefits of a Team OS for you guys?

Dheerja Kaur 23:22 Yeah. Well, so I think one is, like, shared context layer, right?

Aakash Gupta 23:25 Yeah.

Dheerja Kaur 23:25 Like, we all are pulling from the same, or they all, you know, every team is operating a little bit differently, but, you know, they're all pulling from, like, the same data, the same information. I mean, this is, like, table stake stuff, but a lot of the, like, nitty-gritty manual tasks of, like, reporting on data, experiment decisions, right? Uh, flagging and monitoring, like, you know, uh, gaps in what, in, in what's happening in production. Like, all that stuff is happening in, like, one single place, right-

Aakash Gupta 23:49 Very cool

Dheerja Kaur 23:50 ... that everybody can pull from. I think really what we're seeing with the AI team is that there's like, yes, there's a shared context layer, but then there's, like, this workflow perspective of it, like I mentioned, right? Like, you're getting specs checked into Git, right? Engineers are pulling those down. Those can get auto-generated into code at so- in, in some stages, sometimes designs, right? I think every team has a different flavor of that workflow, you know?

Aakash Gupta 24:11 Mm-hmm. Mm-hmm.

Dheerja Kaur 24:11 But I think that's the, that's the dream state, right?

Aakash Gupta 24:14 Very cool.

Dheerja Kaur 24:15 Yeah.

Aakash Gupta 24:15 So what's your take as the leader? Like, are you pushing your teams to embrace all of the latest new AI workflows as soon as they come out? Like, how are you guys properly piloting and figuring out which ones of these things you're gonna take on? 'Cause it feels like there's a new one every single day.

Dheerja Kaur 24:31 Totally.

Aakash Gupta 24:31 Now they're talking about Instinct. It's like, are we gonna all embrace Instinct all of the sudden? So...

Dheerja Kaur 24:36 Yeah. So we- I think we're probably doing what most companies at this stage are doing, which is you can't be prescriptive, right? Because if you say, "Okay, everybody's gonna now use this tool, and this is exactly how we're gonna do it," and then you have to, like, go through change management every single month basically-

Aakash Gupta 24:51 Basically

Dheerja Kaur 24:52 ... if not sooner. Like, you just don't do that. And so we've been pretty explicit with the teams across product engineering, design, even, like, the business teams that we're gonna operate in Wild West mode-

Aakash Gupta 25:02 Mm

Dheerja Kaur 25:02 ... and it's gonna be fun. So here are some tips and tricks and tools. Like, we set up a Git repo for, like, basic Team OS setup, you know?

Aakash Gupta 25:09 Yeah.

Dheerja Kaur 25:09 Something that we actually pulled inspiration from your podcast on. Um, there's also, like, a PM OS version of that for, like, PRD templates and things like that, so people can just sort of get up and running w- they don't have a place to start from.

Aakash Gupta 25:20 Nice.

Dheerja Kaur 25:20 But then make it your own, right? And then have, like, m- tons and tons of forums for people to share what they're doing.

Aakash Gupta 25:27 Mm-hmm.

Dheerja Kaur 25:27 Right? And it's been awesome, right? I think it's, it's been very organic, and what it's felt like is everybody's a builder again and trying stuff, and I think you have to just embrace the Wild West of that.

Aakash Gupta 25:37 Mm.

Dheerja Kaur 25:37 Um, we have been also hesitant to go out and adopt new, like, kind of enterprise-level tools for that reason because I do think that, again, you... Then, 'cause then you have to get into, like, change management, onboarding, and by the time you've done that, is there a new and better tool out there? And so we've let teams, you know, use the t- use the basics, right? Obviously everybody has Claude, you know, et cetera, and kind of build their own tools. [chuckles]

Aakash Gupta 25:59 Very cool.

Dheerja Kaur 25:59 You know? And see how far you can get with that. For example, like, our TPM team built an entire in-house roadmap tool for us-

Aakash Gupta 26:06 Nice

Dheerja Kaur 26:06 ... that auto-pulls from Jira, you know, all of that. And so I think, again, we're... I definitely feel like it's way too early to know what, like, the right processes and tools should be, and you gotta be in this, like, Wild West, figure-it-out mode.

Aakash Gupta 26:17 So it's kind of like I'm hearing, like, three different things-

Dheerja Kaur 26:20 Sure, yeah

Aakash Gupta 26:21 ... in your strategy. Number one, you're actually getting people access.

Dheerja Kaur 26:26 Yes.

Aakash Gupta 26:26 I think a lot of the healthcare PMs-

Dheerja Kaur 26:28 Yes

Aakash Gupta 26:28 ... although you said it's obvious, they actually don't even have access to Claude that I'm talking to.

Dheerja Kaur 26:32 It's true, yeah.

Aakash Gupta 26:32 They're stuck on Microsoft Copilot.

Dheerja Kaur 26:35 Yeah.

Aakash Gupta 26:35 Maybe they have ChatGPT, but they don't even have MCPs or-

Dheerja Kaur 26:38 Yep

Aakash Gupta 26:38 ... API connections.

Dheerja Kaur 26:40 Yep, yep, yep. Totally.

Aakash Gupta 26:40 So give people access. Number two, give people scaffolding if they want.

Dheerja Kaur 26:46 Exactly.

Aakash Gupta 26:46 A starting place-

Dheerja Kaur 26:47 Totally. No cold start

Aakash Gupta 26:48 ... on what-

Dheerja Kaur 26:49 Yeah

Aakash Gupta 26:49 ... we think could be a good thing, and for us it sounds like it's PM OS and Team OS-

Dheerja Kaur 26:53 Yeah

Aakash Gupta 26:53 ... as that scaffolding. And then the third thing, which maybe is the most important, is don't be overly prescriptive.

Dheerja Kaur 26:59 Yes.

Aakash Gupta 27:00 Let your best people discover what the new stuff is-

Dheerja Kaur 27:02 Yes

Aakash Gupta 27:02 ... and then bring that to the rest of the team.

Dheerja Kaur 27:04 100%. Totally, yeah.

Aakash Gupta 27:07 It sounds simple because we just made it three ingredients-

Dheerja Kaur 27:10 [laughs]

Aakash Gupta 27:10 ... but I would guess 90% of teams aren't doing that, right?

Dheerja Kaur 27:14 I always wonder that, if I'm totally honest, 'cause sometimes, I, I mean, as a product leader in this day and age, I, like, worry every night, like, "Are we behind? Are we like-

Aakash Gupta 27:22 Mm-hmm.

Dheerja Kaur 27:23 "Are we at..." You know, everybody has that, right?

Aakash Gupta 27:24 Yeah.

Dheerja Kaur 27:24 Like, the, the anxiety around being AI-pilled enough is real.

Aakash Gupta 27:29 Okay. [laughs]

Dheerja Kaur 27:29 Right? Like, I think that's true. Like, if you talk to folks in the industry. And then of course there's the anxiety around, that everybody has around, like, "I have agents that can do so many things. Am I doing enough," right?

Aakash Gupta 27:38 Yeah.

Dheerja Kaur 27:38 Um, so yeah, I'm, I'm not sure actually. Like, I think at least in tech companies, um, and to your point, I think industry by industry this looks different. I think that this is most of what I'm seeing, you know?

Aakash Gupta 27:48 Mm-hmm.

Dheerja Kaur 27:48 Is that, like, everybody's just building again, figuring it out, and everybody's coming at it with this mentality of, "We don't know what this job is going to look like. We should completely first principles this," you know?

Aakash Gupta 27:59 Mm-hmm.

Dheerja Kaur 27:59 And, like, play around and have fun. And, like, I've said this before, I'm having the time of my life. Like, this is the most fun I've had in my entire career since I was an engineer. [chuckles]

Aakash Gupta 28:07 Mm.

Dheerja Kaur 28:07 You know? Um, and I think that's super fun. Um, I imagine there's a spectrum for sure.

28:12 – 32:17

The CPO job now, and PM vs product builder

Dheerja Kaur 28:12 Yeah.

Aakash Gupta 28:12 At the CPO level-

Dheerja Kaur 28:13 Mm

Aakash Gupta 28:13 ... what's changed? How has your life changed?

Dheerja Kaur 28:15 Ooh, that's a good question. I was talking to my husband about this because in some ways the anxiety is there because a lot of my job hasn't changed, right?

Aakash Gupta 28:23 Yeah.

Dheerja Kaur 28:23 Like, [chuckles] you know, you still have to... L- like a lot of the tactical pieces of it have changed, right? Like, I think a lot of people have, like, chief of staff agents-

Aakash Gupta 28:30 Mm-hmm

Dheerja Kaur 28:30 ... right, that are teeing up, like, even teeing up this, the, the prep for this podcast, you know?

Aakash Gupta 28:35 Yeah.

Dheerja Kaur 28:35 And kind of, like, running down meeting prep and doc creation and all that fun stuff. I also created, like, a Dheerja AI agent to try to codify my product judgment. [chuckles] So, you know, I'm just, like, playing around with the stuff.

Aakash Gupta 28:48 Yeah.

Dheerja Kaur 28:48 But, uh, what I'm finding is, like, that those are more efficiency gains still at, at, for my job, 'cause a lot of my job is, like, running a team, you know, and setting strategy and direction, and then, like, reviewing work and giving feedback and figuring out where the consumer product is going. And so I'm-

Aakash Gupta 29:03 Not to mention hiring a good team too.

Dheerja Kaur 29:05 Yeah. So you know, so it is true that what I'm trying to do is, like, how much of that judgment and kind of role can I, like, continue to codify to, like, really get sort of the step change benefits. But I kind of paused on a lot of the, like, my job aspect of it and was like, "Hey, actually, the mentality I should embrace is that most of my job should actually be about, like, enabling that on the teams," right? Like, that's where you get the most leverage as a business, you know? And so it's less about, like, AI pilling the CPO job and more about my role now shifting to being how do I unlock the benefits of this technology for the teams, and, like, getting the details on that, like, trying to, trying to code stuff myself, right? Trying to, like, play around with different workflows myself, and that's been the fun part.

Aakash Gupta 29:45 Mm.

Dheerja Kaur 29:45 You know? So.

Aakash Gupta 29:46 Coding stuff yourself, and you mentioned being a builder. What's your take on this? LinkedIn, they actually renamed their APM program-

Dheerja Kaur 29:54 Yes

Aakash Gupta 29:54 ... to associate product builder. Recently I had Srini Raghavan on, CPO at Freshworks.

Dheerja Kaur 29:58 Yep.

Aakash Gupta 29:58 He thinks that the PM role might be replaced by the product builder role in the future. Do you think that's likely? Are you embracing the product builder role?

Dheerja Kaur 30:08 You know, it's so funny. This reminds me of, do you remember when Airbnb, I forget how long ago it was, but they were like, "The P- we don't believe in the PM job anymore."

Aakash Gupta 30:16 Right.

Dheerja Kaur 30:17 Right? There's a lot of these, like, hot take narratives.

Aakash Gupta 30:20 Yeah.

Dheerja Kaur 30:20 And I kind of ignore a lot of it because I think it, it really is just this, like, constant renaming exercise-

Aakash Gupta 30:26 Yeah. [chuckles]

Dheerja Kaur 30:26 ... or, like, reinventing of the fact that you still need this fundamental job. What you call it [chuckles] and, like, what you title it, it just doesn't matter.

Aakash Gupta 30:34 Mm.

Dheerja Kaur 30:35 Like, I love that. Product builder, great. I do think product manager, if you really boil it down, like, that's kind of a crappy title 'cause it kind of implies that all we do is just, like, ring lead and we don't actually have, like, judgment or we don't build and we're not hands-on.

Aakash Gupta 30:47 Yeah.

Dheerja Kaur 30:47 But, like, transparently, even pre-AI, the best PMs are, like, builders at heart, right?

Aakash Gupta 30:51 Yeah.

Dheerja Kaur 30:51 So I think this is a little bit of, like, a you get lost in the, the spicy hot takes on, like, LinkedIn and narratives, right?

Aakash Gupta 30:57 Yeah.

Dheerja Kaur 30:57 Um, and so I love it. Great. We can all, all call ourselves product builders now. I don't think that actually changes the fundamentals of the job-

Aakash Gupta 31:03 Mm

Dheerja Kaur 31:03 ... and what you look for. I think what you are seeing, and I'm sure you've seen this a lot when you talk to folks, is that- There are a couple things that are happening with, like, the PM role, or PD or PB role now.

Aakash Gupta 31:13 Mm-hmm.

Dheerja Kaur 31:13 [chuckles] Um, so one is it is causing the folks who spike on the, like, fundamentals. So I'm talking, like, judgment, instinct, taste, all the things that we say are, like, important in AI, and just, like, drive, hustle, ownership. Like, build, get shit done. Those are the things that, like, spike even more than they used to.

Aakash Gupta 31:33 Yeah.

Dheerja Kaur 31:33 But those were the things that most companies that looked for great product people looked for.

Aakash Gupta 31:36 Always.

Dheerja Kaur 31:37 Like, horsepower, right? You know? And then the people who are really, who spike more on just the tactics of the job but don't necessarily spike on that first bucket, I think that, that's, that's the real tension, right?

Aakash Gupta 31:48 Mm-hmm.

Dheerja Kaur 31:48 And the reality is, like, you know, Zurp era, like, hypergrowth in tech, the PM job sort of exploded.

Aakash Gupta 31:54 Yeah.

Dheerja Kaur 31:54 And so you have quite the spectrum of skills. And it is true that there are often, like, different flavors of PMs, and I think those who spike in the fundamentals will c- will just be the most valued people at a company, even though I'm a little biased, obviously.

Aakash Gupta 32:07 Mm-hmm.

Dheerja Kaur 32:08 And then those who are spiking more on, like, the tactics of it, I mean, sure, I think you can reinvent yourself and all of that, but, like, the tactics are getting automated- [chuckles]

Aakash Gupta 32:16 Mm-hmm

Dheerja Kaur 32:16 ... in a way.

32:17 – 43:23

What she hires for and how she interviews

Aakash Gupta 32:17 Briefly covered the fundamentals, but if you had to define them into, like, clear buckets, like these are the three to four skills you should be thinking about, what would those be?

Dheerja Kaur 32:25 Yeah, for sure. Um, and again, I will say this has not changed.

Aakash Gupta 32:29 Mm-hmm.

Dheerja Kaur 32:29 Like, I have looked for this in the last, like, decade of product hiring.

Aakash Gupta 32:32 Mm.

Dheerja Kaur 32:33 And I think it's easy to anchor to, like I said, some of more of, like, the outputs and the tactics and not do enough to, like, interview for the fundamentals, but these fundamentals have not changed about what makes a great PM. So the first is, yeah, like I just said, so just, like, core product intuition, judgment, decision-making, taste. And it depends on what type of role you're talking about, like AI PM versus pure consumer PM versus platform PM versus growth PM, like the, how judgment and decision-making manifest look a little different-

Aakash Gupta 33:00 Mm.

Dheerja Kaur 33:01 Right? But the reality is, like, the highest leverage job for a company is somebody who can, like, make the right calls.

Aakash Gupta 33:06 Mm-hmm.

Dheerja Kaur 33:07 Right? [chuckles] And, like, use their reps that they've learned, use their intuition that they've learned, or just, like, the raw intuition that they have to, like, make the right calls. That's it. Like, that is the number one thing. Second would be, this is w- especially when you apply to, like, this to more junior folks, is just, like, high slope, high horsepower. Like-

Aakash Gupta 33:23 Yeah

Dheerja Kaur 33:23 ... you just know. And you can interview for this, and there are, like, ways to interview for it, but, like, somebody who is just, like, a sponge, super high intellect, high horsepower, high slope. Maybe doesn't have as much of the first sometimes because they just haven't been around the block enough. They don't have the reps. But, like, the second they see something, they pattern match and they pick it up and they apply it the next time. That I will pick over experience any day.

Aakash Gupta 33:44 Over deficiencies in the other areas.

Dheerja Kaur 33:45 You know? Um, and then the third is, like... And these all, these are all soft because, like we were saying, like the tactics of the job are dramatically changing, right?

Aakash Gupta 33:52 Yeah.

Dheerja Kaur 33:52 So, like, I don't, I don't interview for tactics anymore as much-

Aakash Gupta 33:55 Oh, okay

Dheerja Kaur 33:55 ... right? It's like, yes, you understand that, you understand that they can do the job, but, like, these three things are the things that get them to do the job well. The third is just, like, that ownership and drive.

Aakash Gupta 34:03 Oh.

Dheerja Kaur 34:04 You are just, like, hungry, right?

Aakash Gupta 34:05 Being high agency too, right, in that bucket?

Dheerja Kaur 34:07 Oh my God, so high agency, especially when you're in these regulated industries, to be honest. Like, at Robinhood, this was ma- I think the best PMs at Robinhood are the ones who would just be like, "Okay, there's all this, like, ambiguity around, you know, this legal question," for example, right? "I'm gonna go read the law." [chuckles] "And I'm gonna figure this out and partner with legal and just, like, embrace that." And I just think that, like, those have always made the best PMs at every company.

Aakash Gupta 34:30 Mm.

Dheerja Kaur 34:30 So yeah. And then, like, again, the tactics of how you execute are looking so, look so different now than they did, you know, four years ago. And I think really smart people who have those three things can figure that out, you know, and evolve a lot.

Aakash Gupta 34:41 Yeah, they can adapt.

Dheerja Kaur 34:42 So, yeah.

Aakash Gupta 34:42 Especially if they're given the right scaffolding.

Dheerja Kaur 34:45 Yeah.

Aakash Gupta 34:46 So how do you find these PMs now? What's your process look like these days to hire-

Dheerja Kaur 34:52 Yeah

Aakash Gupta 34:52 ... PMs?

Dheerja Kaur 34:52 We do interview loops. We do case studies. You know, we've integrated AI into all of that. I would say [chuckles] I try not to, like, repeat things that I think people have heard a lot. Um, my favorite interview is maybe, like, old school of me, you know? But it's like, "We're gonna spend an hour, and you're gonna take me through, like, the hardest product and problem that you've ever shipped."

Aakash Gupta 35:09 Hmm.

Dheerja Kaur 35:09 "And I need you to take me through every micro piece of it along the way, and we're gonna go really deep."

Aakash Gupta 35:14 Mm-hmm.

Dheerja Kaur 35:14 You know? Um, and we're talking, like, "Okay, wait, this metric moved this, but this didn't, but, like, what about this?" And, like, "What were the four experiments you shipped right after?" And, like, and just, like... Because it's the best pressure test of, like, the can you take a curveball, you know?

Aakash Gupta 35:29 Mm-hmm.

Dheerja Kaur 35:29 And the judgment and the decision-making along the way. 'Cause there are a lot of PMs who can talk the talk in case studies and interviews, right?

Aakash Gupta 35:36 Mm-hmm.

Dheerja Kaur 35:36 Like, there are books on, you know? [chuckles]

Aakash Gupta 35:38 Yeah.

Dheerja Kaur 35:38 There are books on these things, you know? And so, like, really putting them to the test on, like, explain every little decision that you've made-

Aakash Gupta 35:45 Hmm

Dheerja Kaur 35:45 ... along the way and how you corralled and what, like, the gnarly problems that you had to work through. It just picks people apart usually.

Aakash Gupta 35:51 Mm-hmm.

Dheerja Kaur 35:52 And then, like, you know, I think now we're trying to figure out how do you test for people who are also builders, going back to what you were saying earlier, because the way you execute the job just looks different. Like, your classic prototypes before PRDs. Like, are you just ready to get your, like, hands on keyboard to, like, you know, try something versus just, like, write docs and send them over the line? [chuckles] Send them over the fence, you know? And so I think there are ways to do that, and we're playing around with a few different ones. But, like, again, maybe I'm just old school, but I think there are just, like, those fundamentals that are gonna be true no matter what, you know, age we're in.

Aakash Gupta 36:21 Hmm.

Dheerja Kaur 36:22 So...

Aakash Gupta 36:22 Hmm. That's interesting that you mentioned prototypes before PRDs, because-

Dheerja Kaur 36:26 Yeah

Aakash Gupta 36:26 ... I wasn't sure. When we were talking about specs in GitHub earlier-

Dheerja Kaur 36:29 Yeah

Aakash Gupta 36:29 ... I wasn't sure if specs still played a key role, but-

Dheerja Kaur 36:32 Yeah

Aakash Gupta 36:32 ... the prototype is really the main sort of... Because the PRD, what did we used to use it for? Basically, it was, like, a gating document.

Dheerja Kaur 36:42 Yeah.

Aakash Gupta 36:42 I'm gating that legal has signed off on this-

Dheerja Kaur 36:45 Yes

Aakash Gupta 36:45 ... compliance, product marketing, everybody. So is there still a role for something like that?

Dheerja Kaur 36:50 Yeah, it's interesting. So one is the PRD did used to be the beginning of the process, and I think for a lot of those products, you, a picture says a thousand words, right? [chuckles] So to be able to articulate, okay, not just my problem statement, but also, like, here's the thing I'm envisioning-

Aakash Gupta 37:09 Yeah

Dheerja Kaur 37:09 ... is also just more fun. Like, I'm telling, like, what PM would- Prefer to write a doc on that-

Aakash Gupta 37:14 Nobody wants to write a long doc

Dheerja Kaur 37:15 ... to explain it than, like, make something cool, you know?

Aakash Gupta 37:18 Yeah.

Dheerja Kaur 37:18 And I think that happened pretty early on in AI.

Aakash Gupta 37:20 Yeah.

Dheerja Kaur 37:20 Right? Like, I think that was, like, one of the first use cases for PMs, which was like, "Whoa, I can, like, bring this thing to life," and it's so much easier to explain the way I'm thinking about this, and then me and design can jam on this from here.

Aakash Gupta 37:30 Yeah.

Dheerja Kaur 37:30 You know? So that's, like, think has been great. You know what I think that has been around for a while now.

Aakash Gupta 37:33 Just to pause though-

Dheerja Kaur 37:34 Yeah

Aakash Gupta 37:34 ... not everyone is sold on that, if I'm honest.

Dheerja Kaur 37:37 Interesting, really?

Aakash Gupta 37:37 So if you go look at, like, the Itamar Gilads of the world who's like very-

Dheerja Kaur 37:41 Yeah

Aakash Gupta 37:41 ... like evidence-based product management-

Dheerja Kaur 37:43 Yeah

Aakash Gupta 37:44 ... or even Marty Cagan has cautioned against this.

Dheerja Kaur 37:47 Yeah.

Aakash Gupta 37:47 He says if you start building the prototype, you can anchor the solution-

Dheerja Kaur 37:51 Mm

Aakash Gupta 37:51 ... way too soon.

Dheerja Kaur 37:53 Yeah.

Aakash Gupta 37:54 But you guys haven't found that to be a problem.

Dheerja Kaur 37:55 No, I think a lot of that depends on the talent itself and the partnership between product and design.

Aakash Gupta 38:01 Mm-hmm.

Dheerja Kaur 38:02 Assuming you have design.

Aakash Gupta 38:03 Mm-hmm.

Dheerja Kaur 38:03 If you don't, then you gotta do it. [laughs]

Aakash Gupta 38:04 Yeah.

Dheerja Kaur 38:05 Right? Then, then that is it. Because, like, the best designers in the world are the ones who are like, "Okay, I think I understand what you're saying, but I'm not gonna take that at face value."

Aakash Gupta 38:13 Okay.

Dheerja Kaur 38:13 Right? So, like, let me rev on it.

Aakash Gupta 38:15 Mm.

Dheerja Kaur 38:15 And so what we've seen, honestly, we do this amongst the exec team, too.

Aakash Gupta 38:18 Mm.

Dheerja Kaur 38:18 We'll just exchange protos, you know, to like rev. That we had this Slack channel for, like, what does the app look like in the future. And it was just, like, myself, our founder, like, a bunch of, like, ex- We were just, like, literally exchanging Claude protos, like, every day. And it was just fun, and you're just revving on each other. So I think as long as, like, you have the, the people who understand that this is meant to be illustrative-

Aakash Gupta 38:37 Yes

Dheerja Kaur 38:37 ... right? Um, and know how to then say, "Okay, great, I get why, what, where you're coming from, but let me boil that back down and come up with my perspective," it's magical. I could-

Aakash Gupta 38:45 We're not committing to the solution-

Dheerja Kaur 38:46 Yeah

Aakash Gupta 38:46 ... we're really using this to discuss-

Dheerja Kaur 38:49 100%.

Aakash Gupta 38:50 Yeah.

Dheerja Kaur 38:50 Yeah. Um, and but I do agree with that, like, anything, you know, in the wrong context or with not the right, like, dynamics of a working relationship, that can totally, like, stagnate things, right? 'Cause now suddenly you have PMs who are, have to be designers.

Aakash Gupta 39:05 Yeah.

Dheerja Kaur 39:05 [laughs] Um, which I, which isn't necessarily the right skill.

Aakash Gupta 39:07 Yeah.

Dheerja Kaur 39:08 So.

Aakash Gupta 39:09 That's the thing. So we were talking about hiring.

Dheerja Kaur 39:11 Yes.

Aakash Gupta 39:11 And you had mentioned that you guys have case studies and they're integrated with AI. Can you open that up a little bit more? What are your case studies? How are they integrated with AI?

Dheerja Kaur 39:19 Yeah, for sure. So I'm like, should I give the case study prompt? [laughs]

Aakash Gupta 39:23 No.

Dheerja Kaur 39:23 Because we interview on it [laughs] I'm like-

Aakash Gupta 39:25 No

Dheerja Kaur 39:25 ... okay. Yeah.

Aakash Gupta 39:25 But what is the nature of it? Is it a homework assignment?

Dheerja Kaur 39:28 Yeah, for sure. Okay.

Aakash Gupta 39:29 Is it a slide? Is it-

Dheerja Kaur 39:29 Sorry, I was like trying to be like-

Aakash Gupta 39:30 Sometimes-

Dheerja Kaur 39:31 ... I probably shouldn't share our case studies so that-

Aakash Gupta 39:32 Sometimes case studies are just-

Dheerja Kaur 39:34 ... people know how to prep

Aakash Gupta 39:34 ... like if you're in big tech, it's just a 45-minute hypothetical-

Dheerja Kaur 39:37 We do that

Aakash Gupta 39:38 ... that they make you work.

Dheerja Kaur 39:38 Yeah. And this, we did this at Robinhood too. We played around with, like, should we do homework or take-home or not? And transparently, I just, I think people are really busy, especially the best people.

Aakash Gupta 39:48 Yeah.

Dheerja Kaur 39:48 And I just like, I think if you can avoid it, making people do homework, the better.

Aakash Gupta 39:53 Okay.

Dheerja Kaur 39:53 You know? I just do. And I also think that, like, the element of pe- people thinking on their feet s- shows you a lot.

Aakash Gupta 39:59 Okay.

Dheerja Kaur 39:59 I, I really think it does.

Aakash Gupta 40:00 Mm-hmm.

Dheerja Kaur 40:01 And like, sure, maybe you have to sort of like lower your bar for, like, the, the complete amount of output you can get out of, like, an hour of people on their feet versus like, I don't know, a few days or a week at home. But I, I don't know. We, we played around with it a lot at Robinhood, and I just found that, you know, it was just asking a lot for people to take something home, and especially you're often talking to people who have jobs already.

Aakash Gupta 40:20 Yeah.

Dheerja Kaur 40:20 Like, who's got the time for that? And then too, again, like, you don't really know [laughs] how much of that they were able to do and how fast.

Aakash Gupta 40:27 Yeah.

Dheerja Kaur 40:27 So the, the, again, the thinking on your feet is really good, especially for a case. So we do a classic like, honestly, like very similar to what, like, historical, like, big tech interviews.

Aakash Gupta 40:34 Okay.

Dheerja Kaur 40:34 Like classic product case-

Aakash Gupta 40:35 The big tech style product case

Dheerja Kaur 40:35 ... tied to our industry, and like start with a basic prompt, like, "You gotta take me through it from here."

Aakash Gupta 40:40 Mm-hmm.

Dheerja Kaur 40:40 And I think the main AI integration is like, "Show me."

Aakash Gupta 40:43 Okay.

Dheerja Kaur 40:44 And like play around with it, and I think we're, like, batting around different ways to do this, but like, can you actually, like, show me how you use AI to, like, ideate and prototype off of that?

Aakash Gupta 40:52 Okay.

Dheerja Kaur 40:53 So-

Aakash Gupta 40:53 So people will pull up their laptop.

Dheerja Kaur 40:54 Yeah.

Aakash Gupta 40:55 You'll watch them actually use AI.

Dheerja Kaur 40:57 Yeah, yeah.

Aakash Gupta 40:59 Are they using prototyping? How are they prompting it?

Dheerja Kaur 41:02 Yeah.

Aakash Gupta 41:02 Is it a quality prompt?

Dheerja Kaur 41:03 You know, the reason I'm pausing is because I feel like the industry is evolving so quickly that I don't know how good of a signal that is anymore.

Aakash Gupta 41:11 [laughs]

Dheerja Kaur 41:12 'Cause I think that most people at this point are, like, using Claude and stuff to prototype.

Aakash Gupta 41:16 Yeah.

Dheerja Kaur 41:16 Um, but so I think it's... So to me, I don't know how much signal you get out of, out of someone's, like, AI fluency from doing that.

Aakash Gupta 41:23 Mm.

Dheerja Kaur 41:23 However, I do think that, like, a signal of a good PM is somebody who can, to our earlier conversation, like, illustrate their ideas. You know? Um, because suddenly that, the onus is on you to do that. You're not just sending, writing docs and sending them over a wall.

Aakash Gupta 41:35 Mm-hmm.

Dheerja Kaur 41:35 So yeah. But I'm, I'm open to tips and ideas. [laughs]

Aakash Gupta 41:38 [laughs]

Dheerja Kaur 41:38 I think we're, we're figuring it out. [laughs]

Aakash Gupta 41:40 Yeah, I always like to zone in on the hiring component-

Dheerja Kaur 41:42 Yeah

Aakash Gupta 41:42 ... because I think the type of PM that succeeds in today's world may not have been the same as five years ago, right?

Dheerja Kaur 41:52 Mm.

Aakash Gupta 41:52 There's a certain amount of fundamentals that is-

Dheerja Kaur 41:57 Yeah

Aakash Gupta 41:57 ... really, really important. There was-

Dheerja Kaur 41:59 Yes

Aakash Gupta 42:00 ... a certain type of PM, like I interned at Google-

Dheerja Kaur 42:03 Yeah

Aakash Gupta 42:03 ... who I think they didn't even get access to talk to customers [laughs] ever.

Dheerja Kaur 42:08 Yeah.

Aakash Gupta 42:08 You know what I mean?

Dheerja Kaur 42:09 Totally, yeah.

Aakash Gupta 42:09 The corporate came up with the features, so they also-

Dheerja Kaur 42:12 100%

Aakash Gupta 42:12 ... didn't even need the pure strategy. What they were really good at was working within an organization.

Dheerja Kaur 42:18 Yeah, yeah.

Aakash Gupta 42:19 That I feel like has become a little bit less of an archetype-

Dheerja Kaur 42:22 Yes

Aakash Gupta 42:22 ... that can succeed than it could in the past.

Dheerja Kaur 42:25 I agree. I think that, and that's kinda what I was saying earlier about the spectrum and the size of P- how many PMs there are out there has w- it just exploded over the last decade, maybe less. And transparently, what that's meant is that, like I said, you have people who spike on different things, and some of those things might not be as relevant anymore. So I do think that there was definitely, like, there's a long period of time where, like, the pure executor PM, who knows how to, like, corral, [laughs] right? Cross-functional partners. That, that's an incredible skill, you know?

Aakash Gupta 42:59 It had value.

Dheerja Kaur 43:00 Um, it was, it was so much value because, like, you gotta corral, you gotta, like, wrangle the people, you gotta herd the cats, you gotta, like, drive decisions forward, all of that. Now, I think that will still be relevant, but not at the level that, like, you know, it has been. And th- like I said, the judgment, the taste, like, the sort of first principles thinking, the structure thinking, all that stuff. If you don't have that, but you only spec on execution, like, I, I just, I think that's gonna be, that's tough for folks.

Aakash Gupta 43:22 Mm.

43:23 – 52:05

What makes an AI PM different

Dheerja Kaur 43:23 Yeah.

Aakash Gupta 43:23 So you mentioned AI PMs, growth PMs, core PMs.

Dheerja Kaur 43:26 Yeah.

Aakash Gupta 43:26 So you still have a role that is AI PM.

Dheerja Kaur 43:29 Yeah.

Aakash Gupta 43:30 Why is that a different role? Why isn't every PM just an AI PM now?

Dheerja Kaur 43:33 It's pretty, pretty straightforward in that we are actually building AI products. [chuckles]

Aakash Gupta 43:38 Yeah.

Dheerja Kaur 43:38 You know? And so building an AI product- And I think anybody can become an AI PM, for what it's worth. Like, I don't think you need to have some, like, deep experience in AI, 'cause who actually does, right? [chuckles]

Aakash Gupta 43:48 Yeah.

Dheerja Kaur 43:48 You know? Or like ML backgrounds and things like that, right? I think like the best consumer PMs can become AI PMs, but the role does look different, right? Because like the way you build an AI product, like what we just talked about and described, is different from like a classic growth PM that's like, you know, conversion optimization and things like that, right?

Aakash Gupta 44:04 Yeah.

Dheerja Kaur 44:04 Um, and so from a skill set perspective, whether you come into it with this or you learn it, you have to understand the technology at a deeper level and how to work with it-

Aakash Gupta 44:14 Mm-hmm

Dheerja Kaur 44:14 ... th- with those engineers, and it just looks very different than like historically product development looked like.

Aakash Gupta 44:19 Mm.

Dheerja Kaur 44:19 So that's the difference really is like, are you actually building an AI product or are you a PM who uses AI? Which is what should be everybody.

Aakash Gupta 44:25 So building an AI-

Dheerja Kaur 44:26 Mm

Aakash Gupta 44:27 ... product PMs, those are PMs who spike in AI fundamentals and teaming with AI research and engineers.

Dheerja Kaur 44:33 Mm-hmm.

Aakash Gupta 44:34 And they're really good at building AI evals and teaming with the AI eval team. Is that it?

Dheerja Kaur 44:38 At least, at least partnering. Yeah.

Aakash Gupta 44:40 Yeah.

Dheerja Kaur 44:40 I- And I'm sure this is different company to company. You know? I, I'm sure if you work at OpenAI or Anthropic, like the level of depth there is much required-

Aakash Gupta 44:48 Mm-hmm

Dheerja Kaur 44:48 ... is, is higher if you're like a foundational model company. For us, it's a little bit more of this balance of like I need you to have great consumer fundamentals too. [chuckles]

Aakash Gupta 44:56 Mm-hmm.

Dheerja Kaur 44:56 Right? Because you're basically, you're also crafting the experience of interacting with it, right?

Aakash Gupta 45:01 Mm-hmm.

Dheerja Kaur 45:01 But you need to be able to work with the engineering team and like get into the nitty-gritty, 'cause otherwise you can't execute. You're just like crafting a design, and then you're just handing it to engineers, and they have to figure it out. But like the hard part of that job in creating the product that I described we built is not the user experience, [chuckles] right? It's like the evals, like the protocols. Like how do you create a personalization rubric? As we're experimenting with like more proactive outreaches and things like that that the companion can do, how do you build like a conversational guide for those, you know? And so like how you do that is just, it's kind of like retraining the execution muscle, right? And so you have to learn it-

Aakash Gupta 45:34 Mm-hmm

Dheerja Kaur 45:35 ... or you have it coming in, again, which is more rare. And that does... Just like the type of work that you do day to day just looks different.

Aakash Gupta 45:41 Mm.

Dheerja Kaur 45:41 So.

Aakash Gupta 45:42 You mentioned personalization rubric-

Dheerja Kaur 45:43 Yeah

Aakash Gupta 45:43 ... a couple times. I feel like it might mean more to you than me.

Dheerja Kaur 45:46 Fair. Yes.

Aakash Gupta 45:46 Can you break down what is that? What's involved in that?

Dheerja Kaur 45:50 Yeah. And I, I use it as like... So 'cause I like to give concrete examples, otherwise we all talk in like these terms.

Aakash Gupta 45:55 Yeah.

Dheerja Kaur 45:55 So this is one example of many that our AI PM team created as we were creating this companion, and it's kind of what I was referring to earlier, which is we were playing... We were like dogfooding the very first version internally, and it was like clinical only, you know?

Aakash Gupta 46:10 Mm-hmm.

Dheerja Kaur 46:10 Like it would respond to you very clinically. It felt like you were talking to a very impersonal doctor. [chuckles]

Aakash Gupta 46:17 Mm-hmm.

Dheerja Kaur 46:17 You know? So the, a lot of the exact feedback in our product review is like, "Hey, how do we layer on the fact that like we know a lot about you, not just like the, the data, like your clinical data, but also like your goals, and like where you are in your health journey, and who you are as a human? Like we know all these things from these conversations. Um, how do we layer on like tone and brand and the kind of like warmth and conversational style? Um, and then how do we layer on like the synthesis of that information?" And so like to this point of the conversation we're having, that is a classic example of like a skill that an AI PM needs to figure out. Which is-

Aakash Gupta 46:51 Okay

Dheerja Kaur 46:51 ... how do I take that and like codify it in a document-

Aakash Gupta 46:55 Mm-hmm

Dheerja Kaur 46:55 ... that can then be [chuckles] turned into something that actually changes the user experience of interacting with an agent, right?

Aakash Gupta 47:02 Yeah.

Dheerja Kaur 47:02 Which like I don't think that document ever was like a thing, right?

Aakash Gupta 47:07 Yeah.

Dheerja Kaur 47:07 In like classic product development. Like PRDs and specs are much more like straightforward from that perspective. Like-

Aakash Gupta 47:12 Mm-hmm

Dheerja Kaur 47:12 ... you know, if you're like speccing out a UI, you are like, "This button does this, and this button does this," but this is like trying to codify that type of feedback is, is, uh... It's like a, it's just a new muscle. Or that's what I meant by personalization rubric.

Aakash Gupta 47:23 Yeah.

Dheerja Kaur 47:23 It's just like one example of many. Another is what I was saying, which is like we call these things conversational guides, which is our companion proactively reaching out to you or you ask it a question like, "Hey, I'm feeling some side effects." Like there is the clinical protocol, but then there's like, how do I guide the conversation around that?

Aakash Gupta 47:41 Mm.

Dheerja Kaur 47:41 You know? And you have to like c- figure out as a PM how do you codify that for the engineering team to run with. Does that make sense?

Aakash Gupta 47:47 Yeah. What does good look like there? How do you codify that?

Dheerja Kaur 47:49 We're figuring it out, man. [chuckles] Okay, like to really go under the hood, um, it was really funny. The conversational guide, we were just talking about this a few weeks ago. The very first version of these got so prescriptive and deterministic that you... Like, it's so easy to like lose sight of the fact that this is an LLM.

Aakash Gupta 48:07 Oh.

Dheerja Kaur 48:08 You know?

Aakash Gupta 48:08 Yeah.

Dheerja Kaur 48:08 So it's like, [chuckles] you know, just as an example, it's like, okay, if customer asks about this, then start with like this thing, then, then, then do this, and then like, like kind of you craft the response. Then you start, then you get narrower and narrower, and you're like... You know, 'cause, 'cause you start getting anxious, right, as a PM of like, or even like as a designer or whatever. You know, how do I make sure that it doesn't, you know, like it's really like doing the thing that I want it to do?

Aakash Gupta 48:31 Yeah.

Dheerja Kaur 48:31 And so these conversational guides, like I remember at some point the docs got super long. [chuckles]

Aakash Gupta 48:34 Mm-hmm.

Dheerja Kaur 48:34 And I, and then you'd read them, and we'd all be like, "Are we... We're kind of like losing the point here," right? [chuckles]

Aakash Gupta 48:40 Mm.

Dheerja Kaur 48:40 Like, like actually like the whole power of an LLM is that like it's like flexing to the customer itself, right?

Aakash Gupta 48:45 Yeah.

Dheerja Kaur 48:45 And like it's like learning about them. So how do you find this balance between giving it, between like kind of writing that type of like a rubric or a guide so that it's hitting the right points, you know, and not too off-kilter, um, without being like overly deterministic and prescriptive? And like it's crazy, right? It's just like what, like what words do you use? Like how, like what is the structure of that document? Like it's, like we're just like revving on that, you know?

Aakash Gupta 49:06 Mm-hmm.

Dheerja Kaur 49:06 And so it went from this like really, really long deterministic thing to like more of like guidelines. Um, and we're just playing around with it. Like and you had to try it out, right? Like what's working, what's not from that.

Aakash Gupta 49:15 Mm-hmm.

Dheerja Kaur 49:15 So.

Aakash Gupta 49:17 So we talked about this title AI PM. Some people hate this title.

Dheerja Kaur 49:20 Really?

Aakash Gupta 49:20 They're like really against it, you know?

Dheerja Kaur 49:22 Why is that?

Aakash Gupta 49:23 Because they say like it's putting too many roles into one label. Actually-

Dheerja Kaur 49:28 Uh-huh

Aakash Gupta 49:28 ... the head of product at Sarva AI just had a LinkedIn-

Dheerja Kaur 49:30 Yeah

Aakash Gupta 49:30 ... post on it this morning. He said, "Hey, we're putting way too many roles under the label AI PM."

Dheerja Kaur 49:36 Uh-huh.

Aakash Gupta 49:36 "We really need to talk about like-" Maybe you have your AI evals PM.

Dheerja Kaur 49:40 I see.

Aakash Gupta 49:40 Somebody I was just working for just got hired into a purely evals-

Dheerja Kaur 49:44 Yeah

Aakash Gupta 49:44 ... role. Maybe we need to have our AI model PM-

Dheerja Kaur 49:47 Yep

Aakash Gupta 49:47 ... because maybe you decide, "Oh, we should actually be custom-training our models. We shouldn't be-"

Dheerja Kaur 49:52 Yeah

Aakash Gupta 49:52 ... relying. Then maybe you have your AI feature PM-

Dheerja Kaur 49:56 Totally

Aakash Gupta 49:56 ... who's, I think, would probably be owning the personalization rubric-

Dheerja Kaur 49:59 Makes sense. Yeah

Aakash Gupta 49:59 ... that we just covered, right?

Dheerja Kaur 50:00 Yeah.

Aakash Gupta 50:00 So what do you think about that? I know we talked a lot about titles today, but-

Dheerja Kaur 50:04 Yeah

Aakash Gupta 50:04 ... people seem to really zone in-

Dheerja Kaur 50:05 [laughs]

Aakash Gupta 50:05 ... on hating this title. [chuckles]

Dheerja Kaur 50:08 Oh, man, that's interesting. Yeah. I was always somebody who was fundamentally allergic to the overspecialization of PM titles in general 'cause I, like, akin this to... Or this is akin to, like, you know, back in the day when it was like should the... Is growth PM a thing?

Aakash Gupta 50:22 Yeah.

Dheerja Kaur 50:22 Is platform PM a thing?

Aakash Gupta 50:24 I grew up as a growth PM.

Dheerja Kaur 50:25 Right? There you go. And so I think my default mode is like wait until it hits a breaking point where you're seeing such different, uh, profiles of people, and some are clearly successful and some are clearly not-

Aakash Gupta 50:36 Mm. [chuckles]

Dheerja Kaur 50:37 ... for that to warrant, like, this, like, specialization.

Aakash Gupta 50:40 Yeah.

Dheerja Kaur 50:40 You know? And of course, it all depends on, like, what, to what end is this title important? Is it for hiring purposes? Is it for, like, calibrating, you know, what the expectations of a role are? If it's the latter, yeah, I think that you can just get kinda, like, lost in titles, to be honest.

Aakash Gupta 50:52 Yeah.

Dheerja Kaur 50:52 So if I, if I, like, draw that analogy, it's like, okay, there was a time when it was like everybody's a PM, and people just spike on different things. But, like, why are we overspecializing everybody? Because like we said earlier, like, the fundamentals are there no matter where, where you're applying it. But then, you know, I... You, you get to this point where you realize that, like, yeah, the people who are successful in growth versus platform versus, like, zero-to-one consumer, the different spikes are so dramatically different-

Aakash Gupta 51:16 And I feel like they are

Dheerja Kaur 51:16 ... that it's actually... And I, I, I totally agree that they are, that they are... Like, it is very useful from a hiring perspective, from, like, a career development perspective, you know?

Aakash Gupta 51:25 Yeah.

Dheerja Kaur 51:25 And I remember even at Robinhood, we would have people who bounced between them, and it was like a moment. It was like, "Okay, you're now gonna try consumer for the first time. Let's, like, make sure you're learning these skills." So it's, like, very valuable from that perspective. I don't know if we're there yet with AI.

Aakash Gupta 51:39 Yeah.

Dheerja Kaur 51:40 Right? Like, it feels, like, a little early. And fine, let it be the Wild West for now, you know? But it, like... To me, at least for now, maybe depends on where you work, right? Like, I think if you're at, like, a foundational model company, maybe that specialization need is there.

Aakash Gupta 51:52 Yeah.

Dheerja Kaur 51:52 You know? I don't think at our company we're seeing that.

Aakash Gupta 51:54 That's why Sarvam cares-

Dheerja Kaur 51:56 Makes sense

Aakash Gupta 51:56 ... 'cause they're building a foundational model.

Dheerja Kaur 51:57 I totally buy that.

Aakash Gupta 51:58 And so they need the specialization, but-

Dheerja Kaur 52:01 Totally

Aakash Gupta 52:01 ... if you're sitting on top of the models, it's a little bit less important.

Dheerja Kaur 52:04 Exactly.

52:05 – 58:30

How anyone can become an AI PM

Aakash Gupta 52:05 You also said anyone can become an AI PM.

Dheerja Kaur 52:07 Yeah.

Aakash Gupta 52:07 And a lot of people want to-

Dheerja Kaur 52:09 Yeah

Aakash Gupta 52:09 ... not just for the career development-

Dheerja Kaur 52:12 Yeah

Aakash Gupta 52:12 ... and the market signaling. But it seems to be that AI PM roles pay a little better.

Dheerja Kaur 52:16 [laughs] Totally.

Aakash Gupta 52:17 So how does anyone become an AI PM?

Dheerja Kaur 52:19 Yeah. Um, okay, couple things. One, I think you have to be at a company where this is, like, core to the, the bet-

Aakash Gupta 52:30 Mm

Dheerja Kaur 52:30 ... and core to the consumer product if you wanna do it within a company, which is oftentimes the easiest way to tr- to, like, learn something and transition into something. And I think that the reality is that there are gonna be a lot of companies and products where they are looking for their AI feature, but it's just not core to the strategy, no matter how hard you try.

Aakash Gupta 52:49 Mm-hmm.

Dheerja Kaur 52:49 And so I think the most straightforward path is, like, you work at a company similar to what I'm describing where, like, it is the thesis-

Aakash Gupta 52:58 Yeah

Dheerja Kaur 52:58 ... of, like, how that company wins because you're, like, in the thick of it then. And then whether you actually move into, like, a full AI PM role at the company or, like, no matter what you work on, you're getting exposure to how to build AI products-

Aakash Gupta 53:11 Mm-hmm

Dheerja Kaur 53:11 ... like, that is just, like, such a natural, like, graduation, right?

Aakash Gupta 53:14 Yeah.

Dheerja Kaur 53:14 'Cause you gotta, like, get the reps in-

Aakash Gupta 53:16 Yeah

Dheerja Kaur 53:16 ... somehow. And then I think separately, you know, on the flip side, I don't think that there are that many AI PMs out there 'cause it's a whole new world, you know? And so-

Aakash Gupta 53:24 Everyone calls themself a AI PM. [laughs]

Dheerja Kaur 53:26 Yeah. Everybody calls themselves AI PMs. To your point earlier, like, what does that really mean? Is it that you use AI, or are you, like, building an AI product? Like-

Aakash Gupta 53:32 Yeah

Dheerja Kaur 53:32 ... you know, what por- point of that product? So it's like, again, it's, like, kinda the Wild West right now, and I think that at least for us, like, I think a lot of companies do this, like, you just take bets on great PMs, right?

Aakash Gupta 53:40 Yeah.

Dheerja Kaur 53:40 I think it's still early enough that, like, if... I don't think anybody... Maybe not anybody, but, you know, a lot of companies aren't like, "The number one skill set I need is somebody who has done this before," versus somebody who's, like, a great PM with the right instincts around the space that I'm in and high horsepower.

Aakash Gupta 53:56 Mm-hmm.

Dheerja Kaur 53:56 And they, and they have, like, some... They have people to learn from. Um, in our example with our, like, AI PM team, it's a bunch of folks who haven't necessarily worked in, like, AI product, but they're also working with, like, applied AI engineers who are, like, best in class at it.

Aakash Gupta 54:08 Mm-hmm.

Dheerja Kaur 54:08 And so, like, I know they'll learn it from them. And so my advice is, like, oftentimes the best way to learn is, like, adjacency within the company you're in.

Aakash Gupta 54:15 Yeah.

Dheerja Kaur 54:15 And so ch- uh, like, if you're at a company where AI is at its core, like, go for it. Like, it's awesome. We've given a lot of int- we've given internal folks opportunities to do that within our company.

Aakash Gupta 54:24 Yeah.

Dheerja Kaur 54:24 Um, and then alternatively, like, I think people will hire you [chuckles] if you're a great PM, you know? It's still the Wild West out there.

Aakash Gupta 54:31 Yeah. A lot of people listening to that, they're saying, "Well, I can't move internally to my company."

Dheerja Kaur 54:39 Mm.

Aakash Gupta 54:39 "And I maybe haven't been ready to leave either for whatever reason." Is there, like, a path, like a upskilling path that you think could be useful, whether it's side projects or content or something like that-

Dheerja Kaur 54:51 Yeah

Aakash Gupta 54:51 ... in order to get yourself prepared for that next AI PM role?

Dheerja Kaur 54:57 Yeah. But what I'm... What's going through my mind right now is that I think there's so much you can do to learn how to use AI effectively, right? And, you know, reading your, like, newsletter and podcast and, like, you know, playing around with stuff on your own and all of that. I do think to go a next layer deeper to be like, "I can call myself an AI PM," which means that I actually know how to build AI products, that's, like, a meaningful investment, right? That means that, like, maybe on the side, like, you are actually trying to build a product yourself, right?

Aakash Gupta 55:27 That's worthwhile, you think?

Dheerja Kaur 55:28 Yeah. I, I, I mean, right now, like, I think building is so fun that-

Aakash Gupta 55:32 Yeah

Dheerja Kaur 55:32 ... you know, if you have the time for that, like, go for it, you know?

Aakash Gupta 55:35 Yeah.

Dheerja Kaur 55:35 I actually know a lot of people who are, like, taking a mini career break or taking a pause so they can just, like, get in there and learn the technology. So I guess my, my high-level encouragement is, like, to me, it's less about consuming content- And more just, man, if you can carve out, you know, time in your weekday, which for most PMs doesn't exist, to just build hands-on, and then try to go deeper and deeper into it. Not just build, like, the fun, like, you know, I built a little home management assistant or something, you know? But you're going, like, layers deeper and trying to unpeel that.

Aakash Gupta 56:06 Actually have an AI model-

Dheerja Kaur 56:07 Yeah

Aakash Gupta 56:07 ... that you've done something-

Dheerja Kaur 56:09 Totally

Aakash Gupta 56:09 ... whether it's fine-tuned or-

Dheerja Kaur 56:11 Exactly

Aakash Gupta 56:11 ... it's using RAG or something like that.

Dheerja Kaur 56:13 Exactly, yeah. Which is, again, hard to do on your own in, like, a side project, but a lot of people are doing it, you know-

Aakash Gupta 56:18 Yeah

Dheerja Kaur 56:18 ... and I think that could be great

Aakash Gupta 56:19 I mean, it's actually just a few prompts away or a couple end-to-end workflows.

Dheerja Kaur 56:22 Yeah.

Aakash Gupta 56:22 It's easier than it ever was.

Dheerja Kaur 56:24 Exactly. And, you know, I'll, I'll give this story. You know, we have a PM who was on our pharmacy and fulfillment systems. So we build our own software for our pharmacies. In addition to that, was just, like, you know, one of our early adopters of AI and, you know, helped build, like, the PM OS and things like that. And then we just gave him an opportunity to join the AI team 'cause he was like, "I wanna, like, act-" What you were saying, right? "I wanna, like, go multiple layers deep and actually learn how to build AI products." And then, you know, hope it's- hopefully it's okay saying this, but, like, you know, I think, like, a few weeks in, he was like, "Okay, like, yeah, I can do this." Like, [chuckles] you know? Like, it's not that much of a black box, you know?

Aakash Gupta 57:00 Mm-hmm.

Dheerja Kaur 57:00 And so I do think that, like, him even just doing it on his own on his side, like, gave him a lot more priming than he realized.

Aakash Gupta 57:05 Nice.

Dheerja Kaur 57:06 Yeah.

Aakash Gupta 57:06 That's a really concrete example. So what you can do is you can do some side project stuff.

Dheerja Kaur 57:12 Yeah.

Aakash Gupta 57:12 You can try to enable the overall PM team-

Dheerja Kaur 57:15 Yes

Aakash Gupta 57:15 ... with some of the stuff you're learning. That created the confidence in you-

Dheerja Kaur 57:20 Yeah

Aakash Gupta 57:20 ... to then say, "Okay, yeah, you can move into this new team."

Dheerja Kaur 57:23 Yeah, totally.

Aakash Gupta 57:23 And so it's almost about sometimes people think side projects are, like, against your current job.

Dheerja Kaur 57:29 Mm.

Aakash Gupta 57:30 But in a current- in the way this person did it-

Dheerja Kaur 57:32 Yeah

Aakash Gupta 57:32 ... it seems like he framed it in the right way to you.

Dheerja Kaur 57:35 Yeah. And I think that right now in this day and age, learning on your own and, like, having the, like, intellectual curiosity to be like, "I just wanna, like, learn everything I can about this new thing 'cause it's so cool," I think that, you know, at least for me, like, I love leaning into that.

Aakash Gupta 57:52 Yeah.

Dheerja Kaur 57:53 You know? And, like, showcasing that. Whether that means you're moving into an AI PM job or we're just, like, totally elevating that, you know? I think that is, like, that's s- it's just so necessary for the culture that you want at the compa- at companies these days, you know? That, yeah, I think it, it's very different than maybe in the past when side projects were truly side projects. Like, these aren't really side projects. These are, like... One of the best things I think PMs can do on, on our team is, like, have the intellectual curiosity to embrace, like, what their jobs could look like in the future, like, what they could build. And, like, I know that that will pay dividends for their job day-to-day.

Aakash Gupta 58:24 Yeah.

Dheerja Kaur 58:25 You know? And so, um, we have to, like, foster and embrace that for sure.

Aakash Gupta 58:28 Mm.

Dheerja Kaur 58:29 Mm.

58:30 – 1:03:40

The mistake AI PMs make in regulated industries

Aakash Gupta 58:30 So we've had two major threads in this conversation. In the first half, we talked about essentially how to build successful AI products in regulated industries-

Dheerja Kaur 58:39 Yes

Aakash Gupta 58:39 ... like healthcare. In the second half, we talked about how you can become a successful AI PM, whether it's really your role. I wanna kinda end on this note of what are the mistakes that AI PMs in healthcare or regulated industry make-

Dheerja Kaur 58:58 Mm

Aakash Gupta 58:58 ... so that if somebody's listening to this, they can avoid making those mistakes?

Dheerja Kaur 59:02 Yeah, yeah. It's probably going to be pretty obvious. I think what makes a great PM in these regulated industries, even pre-AI, is really understanding and, like, fundamentally intellectually embracing the fact that you are in a regulated place, and whether, you know, in healthcare it's, like, your doctors, in finance it's often, like, your compliance friends, your legal team. They are builders with you.

Aakash Gupta 59:25 Mm-hmm.

Dheerja Kaur 59:25 Like, there's just this mindset you have to have, and I think when you... What I observe for folks not coming from those industries is, like, that is a big learning curve, right? 'Cause oftentimes those are, like, ancillary teams, right?

Aakash Gupta 59:35 Yeah, those are just sign-off teams.

Dheerja Kaur 59:36 That are, like... They're just sign-offs. Exactly. They're just sign-offs in a, you know, PRD, right?

Aakash Gupta 59:39 Yeah.

Dheerja Kaur 59:40 Um, it's so different in these spaces. Like, you have to build a solution together.

Aakash Gupta 59:44 Mm.

Dheerja Kaur 59:44 You just do. And so when you're... Because oftentimes, like, half the battle of building these products is understanding the constraints-

Aakash Gupta 59:51 Yeah

Dheerja Kaur 59:51 ... and then figuring out the creative solutions to do so. And what's awesome is that in the best companies in these industries, they've hired folks in these roles who are, like, creative solutions thinkers, right?

Aakash Gupta 1:00:02 Yeah.

Dheerja Kaur 1:00:02 Like, our... I mean, I'm obviously biased, but, like, our Hims and Hers legal team is, like, world-class, right? Like, they're so good at what they do. They know their stuff, and they, like, understand what we are trying to do in the industry and, like, know how to partner with, like, builders to do so. And that's incredible, but you have to, like, r- totally rethink your relationship with those functions compared to other industries.

Aakash Gupta 1:00:22 Yeah.

Dheerja Kaur 1:00:22 Right? And so going back to, like, the point of your question, yeah, so one is, like, you know, embracing that and not, like... And, and, and checking your biases-

Aakash Gupta 1:00:29 Yeah

Dheerja Kaur 1:00:29 ... on how to work with those functions. And then, like, just [chuckles] recognizing, like, the magnitude of what you're doing. So if I go back to the example of what we've shipped, we've launched an agent that is the, like, front door of the consumer experience that is directly autonomously interacting with customers. That's awesome, and that also comes with, like, real, like, risk. [chuckles]

Aakash Gupta 1:00:48 Huge.

Dheerja Kaur 1:00:49 Right? And so it's not just, like, a one-time solutions thing. It's, like, an ongoing engagement, right? We have our legal and clinical partners in the room every day in a standup, right?

Aakash Gupta 1:00:58 Mm.

Dheerja Kaur 1:00:58 Like, looking at what's going on in production and, like, monitoring-

Aakash Gupta 1:01:01 Wow

Dheerja Kaur 1:01:01 ... and all of that, and there's just, like, a MO of how you operate. Um, and if you don't, you might just have blind spots and you make, make, make, make mistakes, and, like, that's also my job to make sure that doesn't happen, you know? Um-

Aakash Gupta 1:01:11 Yeah

Dheerja Kaur 1:01:11 ... but I think it's also, like, inherent to the, to the PM role.

Aakash Gupta 1:01:13 All right, guys. We just walked you through a live case study of a company that had a very successful product. This is a public company. They went through the agentic transformation. They put it on the front door of the product in a regulated industry. This is going to be what all companies end up doing in the future. You just got a taste of it, what the role looks like. Dheerja said this off camera, but their entire team is using the Team OS that we published in this newsletter on this podcast. Go pick one of those up. Like she said, go check it out in your side time. You don't necessarily need to wait for your company to give you the scaffolding. You're one of those 1% of people who are going way above and beyond and listening to the end of this podcast. So like she said, now shift from the content to doing. Go send a couple prompts over to Claude Code. Here's my instruction to you. Claude Code, let's build an AI product together. I want it to have RAG and fine-tuning. I'm passionate about X. Where should we start? It can literally help you from that simple prompt. So go learn the tools, go get the job, go apply to Hims & Hers, go apply to a regulated industry. As you see, they're actually embracing AI totally in a way that surprised me. And good luck with your career. We'll see you in the next episode. I hope you learned as much from today's episode as I did. If you can do one thing that's totally free that would help the show, it would be to check that you're following on Apple and Spotify podcasts. Check that you've left ratings and reviews on those platforms. Check that you're subscribed on YouTube. Leave a like and a comment on this video, and then share it with your friends. We're trying to make better and better podcasts. After two years, we think we've gotten something pretty good going, so let us know what we can do to make it even better, who else we should interview, and we will put on the best shows we possibly can. Finally, don't forget my offer for the bundle. You get an entire year of my paid newsletter, plus my favorite AI tools, Bolt.new, Airtable, Speechify, Descript, Magic Patterns, Linear, Dovetail, Arise, and Mobbin. That's $27,000 worth of value for just $150. So check that out at bundle.aakashg.com if it interests you, and I can't wait to share our next episode soon.

Episode duration: 1:03:50

In the uListen app, the transcript lights up word by word as you listen. Get uListen

Transcript of episode j1jRqOLHl_w

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.