1% better

The new rules of product management in the AI era | Sherif Mansour (Head of AI, Atlassian)

What does a product manager actually do when AI can build nearly anything? Sharif Mansour, Head of AI and Product Craft at Atlassian, argues that the core...

1h 21m

Summary published by , updated .

Lenny's Podcast

Key Takeaway

Identify the context gap your team has before reaching for an AI tool.

Before your next sprint planning or strategy session, ask: what do my engineers or designers not understand that is causing us to make slower or worse decisions? Then find a way to close that gap directly, whether by bringing an engineer into a customer call, running a shared market analysis session, or building a vision together.

Sharif Mansour's argument is that throughput is becoming cheap and abundant, so the scarce resource is direction. Getting the right people to understand the right context produces better decisions across the whole team, faster than any individual could manage alone.

Episode Overview

What does a product manager actually do when AI can build nearly anything?

Sharif Mansour, Head of AI and Product Craft at Atlassian, argues that the core job has always been filling context gaps, and that this matters more now, not less. He draws on 17 years at Atlassian, including hands-on experiments with AI builder teams, to explain why PM hiring is growing while other roles stagnate.

The conversation covers three main areas:

  • Context as the job. The best PMs build a shared brain across their team rather than making decisions for them.
  • PM archetypes. A triangle framework of artist, scientist, and general manager types helps explain why PMs thrive or struggle in certain roles.
  • AI-era hiring. Taste, cultural openness, and AI-native thinking are rising in value, and junior hires bring unexpected leverage to established teams.

Main Insights

1. A PM's job is filling context gaps, not making decisions

Many PMs measure themselves by how many decisions they make or how much they control the backlog. Mansour's view is that this misses the point entirely.

He describes the PM role as building a shared brain: making sure engineers, designers, and other teammates understand the customer, the market, and the business well enough to make good micro-decisions on their own. When that context is shared, the team moves faster without waiting for the PM to approve every step.

A practical starting point: before you decide something, ask yourself why you think one option is better. Then ask whether your teammates have the same information. If they don't, closing that gap is probably more valuable than making the decision yourself.

Mansour also suggests never doing a customer interview without an engineer in the room. The empathy built from direct observation is far more durable than a handed-over summary.

2. AI increases throughput; direction becomes the scarce resource

When Atlassian experimented with squads made up entirely of AI builders, with no dedicated PMs or designers, the early weeks felt fast. Within a month, those teams had quietly returned to their old roles.

The PM on that team explained it plainly: throughput isn't the problem; direction is. AI makes it cheaper to build things, which creates more prototypes, more ideas, and more noise. What teams need is someone helping them decide what to build and why.

This is why PM hiring is rising while other roles plateau. As building becomes easier, the value of steering the ship grows. Mansour's internal post at the time was titled "Direct the ship, don't row it."

3. Match the PM archetype to the kind of work, not just the opening

PMs are not interchangeable. The PM craft triangle, developed by Geoff Redfin at Atlassian, describes three natural orientations:

  • Artist: vision-led, experience-focused, more data-informed than data-driven, tends to sketch or prototype first.
  • Scientist: hypothesis-driven, comfortable with A/B tests and growth metrics, starts with a question and works toward evidence.
  • General manager: thinks in TAM, revenue, and business model; frames problems in terms of market opportunity.

Mansour says you can often tell where someone sits just by asking what they would do to improve a favorite product. The answers look completely different across the three types.

The practical implication: when a PM is struggling, the problem may be a mismatch between their natural orientation and the kind of work the role requires. Moving them to a better-suited project often produces an immediate shift.

4. Hire junior people to bring AI-native ways of working into established teams

Many organizations assume that AI can now do what junior employees used to do, so there is no reason to hire them. Mansour argues the opposite.

Juniors who have only ever worked with AI tools bring new habits that established teams find hard to learn. They are more likely to try different tools, experiment with new workflows, and treat AI-assisted work as the default. Seniors, by contrast, are better at spotting AI slop because they know the domain well enough to recognize errors.

The combination is what creates value. Juniors surface new possibilities; seniors catch the mistakes. Atlassian's internal AI Builder Weeks make this visible: younger participants tend to run demos and find novel tools, while senior staff focus on evals and quality problems.

Mansour cites internal and customer research showing juniors are about 38% more likely to adopt new tools than seniors.

5. Find the work you are naturally good at and build a complementary team around your gaps

There is a common assumption, especially in product, that good PMs should be strong across all three archetypes. Mansour's experience, including his own career, suggests this is rare and probably not worth chasing for most people.

He describes himself as a strong artist PM who is openly poor at the general manager mode. Rather than trying to close that gap, he works alongside a partner who is strong where he is weak. The public acknowledgment of his own weaknesses, he says, makes it safer for others to do the same.

For people earlier in their career, his suggestion is straightforward: try many types of projects to discover what you are naturally good at, then look for the signal that you are good at something by noticing when others respond well to your work. From there, seek out more of that kind of work and tell your manager explicitly.

Frameworks or Models

PM Craft Triangle

A model for understanding how different PMs approach product discovery, created by Geoff Redfin at Atlassian. Each corner of the triangle represents a natural orientation:

  • Artist: leads with vision and experience, tends to start with a solution or prototype, is more data-informed than data-driven, and relies on intuition and taste.
  • Scientist: starts with a hypothesis, relies on A/B testing and quantitative analysis, and is comfortable in growth-focused roles.
  • General manager: frames problems in terms of market size, revenue, and business model, and thinks about where to play and how to win.

Most people sit closer to one corner than the others, though some can operate at a basic level across all three. The framework is most useful for matching PMs to the right kind of work and for building teams with complementary strengths. You can often identify someone's orientation from a simple question: what would you do to improve your favorite product?

Three types of context

Mansour describes the PM's job as filling three recurring types of context gaps within a team:

  1. Customer context: what customers need, how they behave, and what problems they are trying to solve. Best built through shared experiences like joint customer interviews.
  2. Product and market context: how the product fits the competitive landscape, what alternatives exist, and where the market is heading. Can be built through shared visioning and landscape analysis.
  3. Business context: the goals, constraints, and priorities of the organization the team operates within.

The PM's job is to identify which type of context gap is causing the team to make slower or worse decisions and then find activities that close it. As teams scale beyond 20 to 30 people, these gaps tend to widen and require more deliberate effort to maintain.

Notable Quotes

"We realized throughput's not our problem. Direction's our problem."

Sharif Mansour

"AI is really good at amplifying your strengths and equally it amplifies your weaknesses."

Sharif Mansour

"Because you have access to the tool doesn't mean you're an expert at the domain that you have access to."

Sharif Mansour

Action Items

Suggested exercises inspired by the conversation.

  • 1
    Audit your team's context gaps before your next planning cycle

    Before your next sprint or quarterly planning session, write down three questions:

    1. What do my engineers not understand about our customers that is slowing decisions?
    2. What market or competitive information does the team lack?
    3. Where are people waiting for me to decide something they could decide themselves?

    For each gap, identify one activity that would close it, such as a shared customer call, a market review session, or a team vision workshop. This reframes planning from task prioritization to context distribution.

  • 2
    Bring an engineer to your next customer interview

    The next time you schedule a customer call or usability session, invite one engineer to join as an observer.

    • Brief them beforehand on what you are trying to learn.
    • Let them listen without taking over the interview.
    • Debrief together immediately after.

    Mansour argues that the aha moments an engineer builds from direct observation are far more durable than a written summary. Notice whether their questions or suggestions in the following week reflect that exposure.

  • 3
    Map yourself on the PM archetype triangle

    Draw or sketch a triangle with artist, scientist, and general manager at each corner. Place a dot where you think you naturally sit based on how you approach product discovery.

    Then consider:

    • What kind of work has produced the best results and the most positive feedback in your career?
    • Does your current role match that orientation?
    • Where are the gaps in your team, and who might fill them?

    Share this with your manager as a starting point for a career conversation about the kinds of projects you want more of.

  • 4
    Practice arguing against your own recommendation

    Before your next product review or decision meeting, write down the strongest case against your own proposal.

    1. List two or three reasons the opposite approach could be right.
    2. Present both arguments in the meeting before stating your recommendation.
    3. Ask: why would we not do this?

    Mansour says the best PMs can intellectually wrestle with the opposite view and still move forward with conviction. This also makes it easier for others to raise concerns they might otherwise suppress.

Go deeper

Full Transcript

Read or copy the complete episode transcript.

When AI became a thing, many people were like, PMs are over. We don't need them. For a while, especially a year ago, it felt like the narrative was everyone's becoming an AI builder. We moved some of our product managers and designers to a team. Everyone was the AI builder. There was no PM or designer. I remember checking in with one of the PMs in the first few weeks. And he's like, it's unbelievable, man. We've moved faster than ever before. After about a month, he comes back and he's like, everyone's gone back to doing their traditional jobs. We realized throughput's not our problem. Direction's our problem. I need more people to steer the ship. In spite of what we keep hearing, PMs should be shipping code. PMs should be building. PMs are some of the worst solution designers in the world. Because you have access to the tool doesn't mean you're an expert at the domain that you have access to. AI is really good at amplifying your strengths and equally it amplifies your weaknesses. You spend a lot of time thinking about what makes a great product manager. They don't own decision-making. PMs own decision-making velocity. The largest, most important part of their job is they're literally filling context. The best PMs are the ones that are thinking about building a shared brain at any scale, product or software team. What's trending up in what you are looking for when you are hiring PMs? This may sound a bit spicy. One misconception I'm often debunking is you don't need to hire juniors anymore. They bring different ways of working to the team. My 12-year-old son vibe codes his own games. He's gone from scratch straight to vibe coding. He's missed the whole syntax and whatever. He thinks that's normal. Sharif, thank you so much for being here. Welcome to the podcast. It's so awesome to be here. Thank you for having me. A long-time listener, first-time caller, so it feels surreal, but good to be here. I get that a lot. You are head of AI and product craft at Atlassian. You and your team build some of the most important and influential products in the field of product. Jira, ever heard of it? Something that is very cool about you and your job is you spend a lot of time thinking about what makes a great product manager, how to develop great product managers, how to hire product managers. You spend a lot of time on the craft and the skill of being a product manager. Let me ask just this big question and see where it goes. What makes a great product manager? In any medium to large organization, I'm going to focus on that because I think it's quite different to a startup. I think that the largest, most important part of their job, to use a phrase that everyone loves using today, context, is I think they're literally filling context. They're filling context with humans and I guess these days with agents too, but they're the ones who realize that the best way I can advance this product is not by me taking all the decisions and all that stuff, it's one of my popular misconceptions I love to debunk, product managers and all the decisions. But it's about if they can build a shared brain with their team, their cross-functional team, especially their engineers, I always say three types of context to my teams. You have the customer context, you have the product and market context, and you have the business that you operate in context, and you're always filling context gaps, like the hardest thing I've had to do, Lenny, if someone asked me the other day, what's the hardest thing you've ever done at Atlassian? I feel like without a doubt, it's just scaling product teams. Everything breaks, trust breaks, shared brains break, us and them culture starts forming, in any scaled organization, more than 20, 30 people, you're going to run into this. So having a job of building a shared brain as a group on those three contexts becomes key because if you can empower everyone with that context, they can move faster. I always say to PMs, your job is to stay away from the Jira backlog. And they're like, wait, you work at Atlassian, you can't say that. And I'm like, no, no, hang on a second, hear me out for a sec. Run away from your backlog because why is it that you think some card is more important than another card? There's some context that you have that the rest of your team doesn't. So how can you power your product engineers, your feature leads, your designers, et cetera, with that context so that they can make those decisions themselves? Now, granted, you may never work yourself out of a job, but you're always filling context gaps. It might be a vision gap this year, next quarter might be a customer gap, might be a landscape gap. Like you're always trying to fill context. I think the best PMs are the ones that are thinking about building a shared brain at any scale, product or software team. So a big implication of this advice, because I don't think many people would think this is what makes a great PM. This is a really interesting and unique take, this idea of just providing context for the team versus, you even said this, like your job as a PM is not to make the decisions, it's to help your team make better decisions. And through that, it's giving them context. So kind of the implication here is your job is to help the team have all of the information such that they can all make the right decision versus you having to convince them or be the decision maker. Yeah. I remember early on in my career, I remember sitting down with my manager saying, like, this is insane. I have so many decisions I need to make. The team's blocked on me. I need to move in this direction, in that direction, in that direction. And she looks at me after me venting for like 10 minutes straight in the eye. And I could tell she was sort of looking past me, not really listening. And she says, Sharif, you just need to spend more time on the strategic things. And I'm like, how do you have time for these things? I'm too busy planning the sprint, working on what we're doing next, et cetera. And you realize that the best way to make you and the team scale is by making sure that everyone has a shared brain. You know, it can go by simple examples. Like, I tried never to do customer interviews without an engineer in the room. Like, you're building context in your brain. Like, sure, I can do the interview, then get an agent to summarize all that stuff and hand it off to a human and say, look, here's the findings. But if you're sitting in the same call as me, Lenny, like your aha moments and your empathy that you built for the customer is, you know, a hundred times better than someone who's just handing over reports of, here's my customer findings. So, you know, that's an example of filling customer context. How can I do that in a way that more engineers feel empowered for that? But, you know, product and strategy context, you might get your team involved in visioning together and looking at market landscape together and trying out different tools together. And rather than one person doing it and, you know, coming down from the mountain and handing over these instructions. This episode is brought to you by this season's presenting sponsor, WorkOS. What do OpenAI, Anthropic, Cursor, Repl.it, Sierra, Clay, and hundreds of other winning companies have in common? They're all powered by WorkOS. If you're building for enterprises, you have felt the pain of integrating single sign-on, SCIM, RBAC, and audit logs. WorkOS turns those deal blockers into drop-in APIs built for B2B SaaS. And now there's a new kind of user that needs access to your app, agents. AI agents need to register, authenticate, and get permissions, just like a person. WorkOS handles it all. Agent auth to give agents scope for vocable identities, agent registration with auth MD, and airlock to judge every agent action and allow, deny, or escalate to a human. I've noticed that every startup that I'm an investor in that expands up market ends up using WorkOS because they're the best. Go to workos.com to make your app enterprise ready today. I'd love to hear more examples of what this looks like in practice of either what not to do, to do this badly, and what you think are kind of counter examples of a PM and, or doing this context sharing really effectively. If you think something's more important than something else, ask yourself, why did you conclude that? And like, you know, reverse engineer your thinking and what made you realize that? And why does the other person that you're telling them that does, why do they know that? And so it might be because they don't have the data points of the quantitative or qualitative data that went involved there, et cetera. It might be because, you know, they went in those meetings or they don't have that information. And so, you're always thinking about what can I do to bring people along on the journey? So the best designers, as a great example, aren't handing over the designs necessarily. They're getting the team to co-create together, and they're exploring concepts together. Now they may take that all and then finesse it and finish it and bring out this, but when the team's involved in that process, their aha moments completely change. Another example might be, I had a product manager the other day that's getting every engineer in the team to go and use different AI tools and to come back and tell them what worked and what didn't work for them. And that's something we typically wouldn't do. Like we'd say, oh, the PMs or the designers will go do that, but like you're getting everyone involved in the process is a huge one. We just came out of our team, our biannual event in Europe with customers. We try to take engineers to the booths and engineers man the booths with me and other people, and they're fronting customers and hearing feedback. And that's ways we can build shared brains together, right? So a lot of that. Now, granted, you may start every project and you take the decision yourself and you're trying to get things going, but your job always as a PM is like, how do I move away from this thing? How do I get better alignment on the vision? They put this on the roadmap and I disagree with it. Why do I disagree with it? Then maybe there's some strategic beliefs that I have that I haven't built a shared brain around with the team, or maybe I'm wrong and they're right. And so you're always trying to close those gaps. So you're always trying to find where are the gaps and then what activities can you do to close those gaps? So interesting. So there's kind of this spectrum that you could think about PMs of how they operate at different companies. And one of the spectrum, they're the boss, make all the decisions. They are basically managers of the team. The other is they're just completely equal, no additional power of any kind in decision-making. Where in that spectrum do you think PMs should be? They're always striving to be in the equal, no additional power. Now think about it also, in theory, PMs don't have, a PM working in a squad has no reports to them. They have to influence without authority, right? By definition. So, and there are two operating models here. There is the, I will hand over the spec and you will do what the spec says. Or there's the model of, I'm going to get you all bought into the vision and we're going to work on this together. To the point where the best engineers are almost like suggesting their own features to solve the customer problems. Some of the best stuff that we've had, if I could share a quick example, I remember, I love telling this story because it's an old story and people are like, what? In 2010, I remember interviewing a set of Confluence customers and we could see that before emoji even existed, customers were putting in these little icons in their Confluence pages, right? And now quantitatively, you've got the data and you're like, oh, people are putting in these icons, we should ship more icons, give them more variety of icons, whatever, and we particularly found in this one customer call that, you know, they would put these red stars and these green stars in their meeting notes in Confluence. And I'm like, and we saw this with multiple customers, like a different system they built and we're like, why are you doing that? And, you know, they would be like, oh, the red stars are the things we haven't done and the green stars are the things we've done. And this engineer at the time left the meeting and she's like, why don't we just put a checkbox in a Confluence page and let them like actually take it. And don't forget, this is 2010, no one had done this on the web, like putting tasks in a document editor was not a concept that existed. And she just pioneered that because of her aha moment from observing the customer. We wouldn't have got that otherwise. Look, there's lots of examples of these, but when they're involved in that journey, you're making them the mini product managers in each of the areas. I know you also have this framework of how you think about archetypes of PMs. There's artist PMs, there's a couple others. Talk about that, because I think a lot of people, they hear PM and it's like one type of person. And it turns out you can be a successful PM in different ways. Talk about this framework. Yeah, we have a framework that is, Geoff Redfin actually created it when he joined us. He shared it with me and we're stuck with it then. You know, there are different types of PM frameworks out there. I think if you pick one and stick with one, it's useful. But he has what we call, he calls the PM craft triangle that, you know, one side of the triangle, he has a general manager. Other side, he has a scientist and the other side, he has an artist. And the positions of the triangle represent how you might do product discovery more naturally. So as an example, a scientist PM will be very hypothesis driven. Think of them in growth teams. Think of them as A-B testing a lot, etc. General manager PM will think more in TAM and business and profit and revenue and, you know, all that kind of stuff. And an artist PM will be more just emotionally vision led and experience led. And I found this to be a helpful framework because the classic, let's just throw a PM in a team isn't the right thing to do. It's like, don't throw bodies at the problem. Think about like, what type of PM are we after and how can we best map that PM? So we use that a lot in calibration when we're comparing PMs. We use that a lot in hiring. Like you'll, you'll ask a general question, like what's your favorite product and what will you do to make it better? It's, you can tell someone's where they sit in the triangle by how they answer it. Like the make it better answers for the general managers, often like there's a new business opportunity we should go after. And I think the TAM is this big and I'm going to measure it and whatever. And the scientists will start with a hypothesis and they'll be like, Oh, we could, you know, test these different variants, et cetera, and the artists will just paint some visual vision and then kind of back their way into what the problems are, but they'll start with like this solution experience that they want to paint because they feel emotionally led towards that destination for a particular reason, their intuition, their usage or whatever, and they'll walk backwards and so you can do that. So it's a, it's a very helpful tool to say, Oh, we need a new PM on this team. And I'm like, Oh, what kind of work are we looking at? And do we need that more of an artist, a scientist, general manager? You get some PMs that say, I'd like to do all three. The reality is I found very few people in my career that can do all three really well. Uh, and, but you can get a lot of people to do all three at a, at a basic level. And they're, they're awesome for zero to one projects, startups, et cetera. Like, as in you need to do whatever you need to do to get to the outcome that you need that cultural law in smaller zero one projects or startup teams. Uh, the bigger you get, the more you can, you have the luxury of being like, we need someone just to really focus on the growth metrics of this particular product in this particular area or et cetera. What I love about this, there's a bunch of reasons. One is it shows you, you can be successful in different ways and you don't have to be one specific type of PM. Two is this explains why you may be suffering and failing right now is you may be the wrong kind of PM for the team you're on now. Lenny, the amount of times I've had someone who may not be finding their groove on a particular project, swapping the type of project they're on and they've bounced back, got a new found energy. I can point to so many examples that it happens all the time. And it's a classic of like, just find the work that's most suitable to that person. And just, it's not about just throwing PM bodies at the problem. Uh, it's just about who would have thought diverse teams still matter. What kind of person do we need in this team? Uh, how do they compliment the other people? What's the kind of work and. you know, you kind of map that day. I know that you also have this, and I share this philosophy that you don't need to be great at everything, just as an employee and as a PM as well, that you don't need to be great at everything. Instead, figure out what you're strong at and focus on that. And it connects to this framework. Talk about just that, and I think it's just this kind of myth you think a lot of people believe. Yeah, it's also, selfishly Lenny, it makes me feel good. So the running joke in Atlassian is, I'm like the worst general manager type person to ever have. Like, if it's about running the business, don't give that to Sharif. Now, I'm in the exec team, I'm the head of AI, I run the product craft, and I love that people can joke about that about me, because I think what that does Lenny is it creates a safe space for other people to be okay to share their weaknesses. And then here's the challenge people run into when they run into a weakness. They have two options. They could try to get better at that thing. Now, sometimes that's left brain, right brain stuff, and they might need to work harder at one thing than it is nature versus nature kind of topic. Or what I have found to be more useful in my career, and I've tried to be the, oh, I'll get better at that thing, is actually find the thing you're good at and run for the hills towards that direction and ignore everything else. And then build a diverse team around you that are good at the other things. So I am a very heavy, strong artist to PM, but in my, I'm responsible for AI across Atlassian, I have a compliment, Jamil, he works with me. He is the complete opposite of me, and we are such a good team together, and I'm so grateful for him, because in my weaknesses he shines and his weaknesses I shine, but together we're an awesome team. And so I always say, if you're early on in career, try as many things as you can to discover, we discover products, we do product discovery, you should do career discovery, discover what you're naturally good at. Then you have two options, balance it out, or just keep running towards the thing you're naturally good at. My recommendation's always, run to the thing towards you're naturally good at, celebrate your weaknesses publicly so that you can build a complimentary team around you and create a safe space for everyone else to feel like they can share the same. I believe it's important to say, you can't be terrible at things, like there's probably a baseline, you have to be good enough at meetings, good enough at strategy, good enough at taste, but you don't need to be the best at every category. Correct, yeah, I think it's the, again, one of those misconceptions that everyone wants to be Steve Jobs-esque, like I'm good at everything, et cetera. Now, Lenny, don't get me wrong, there are a couple people in the company, like I would use Mike, our CEO, co-founder as an example, he will be in a strategy review with you, and he'll tell you why your drop shadows are off by five pixels, why the padding and margins are off on the left here, then he'll go tell you why your business model is completely wrong and flip it around, and then he'll go and look at the data inside out and tell you something you had no idea about. Those people are incredibly hard to find, but the vast majority of folks, I think, find the area they're good at and just keep going in that area. Let me ask you this, when you coach people in their early career and then in later career, if you had to pick one or two most common unlocks that help them accelerate, so say you're coaching someone early career, if you think about the people that you've helped at the beginning of their career, and then the second part is just later career, what are kind of the most common unlocks that allow them to all of a sudden become a lot more successful at product management? Giving them an opportunity that is just above their level at the right time is key. And that opportunity, you should be very deliberate about, it's something that plays to their strengths specifically. For example, this product manager comes to mind that she used to work with me in the early days of a new product that we launched a few years ago at Atlassian. And I realized at the time that she was at the stage where she was more in the scientist brain that the product was growing. And I realized also that like, I'm probably not the best to lead this. And I stepped aside and just said, hey, you know, Rachel, you should probably take over this piece of work. So I think it's just really about finding, if you know where each person is at on their journey, and you can see where they want to head and what they're good at, waiting for the right opportunity, your job as a leader is to see that opportunity arise, and then that might be moving them across to a different team or whatever it is. I think that is the sweet spot of giving people something. I've been fortunate to be here 17 years. And like every time I've looked around, there's a new opportunity at Atlassian. This is like, because we've just grown so much. And these opportunities have happened to play to my strengths. So, you know, in the same token, I need to be looking at that for the people in my team and passing that on as well. So I think that's key is just identifying opportunities inside the business is such an important thing that doesn't get talked about as much, yeah. So say early career PM is listening to this, they're gonna be wondering, okay, what kind of opportunity should I be looking for that is most likely to help my career? What are signs that this is a good opportunity that may accelerate their career? It's probably started with someone telling you you did a good job at something. That's probably a good sign that that's the type of work that you're good at. So if you're feeling like you've found that and people saying that, you know, they really like the thing I did here, this particular project, they really love how I led growth in this particular area, or they love how, you know, I solved this customer problem in this area. Analyze that type of project, then ask yourself, how did I approach it? How did it work? Then your next thing is, what other types of projects that are similar that I can go do is what you should be looking at, tell your manager about. I always say to folks, a good conversation to have your manager is pull up one of those frameworks, like here's where I think I am in the triangle, whatever framework you want, there's lots of them out there. I'd love to do more of this stuff if you're still discovering, or I would love to get better at the thing I'm good at here. Put it, you know, tell them, indicate to them that you're after the next opportunity I think is a huge step you can do to help yourself. What about, let's say a senior PM, you know, I don't know, like L6, or even like early manager, what's kind of the most common unlock that accelerates their career from that point? I think at L6, you've really found your groove, like as in, so I think you're at that decision where maybe you don't want to accelerate, if you say accelerate your career, Lenny, and you mean accelerate your levels, maybe you want to go diverse, as in go try something totally different. You could argue, is that accelerating career? I would still say that's accelerating your career, right? So, or just keep chasing the similar type of work. I think it's key. You know, this particular lady in one of my teams yesterday, we celebrated her 10 year anniversary at Atlassian. She found the kind of work that she's really good at. She leads a lot of our pricing and packaging work across AI and all that kind of stuff. And she's so good at it. And we just, she keeps finding more work across the business that lines up with that. So she's, you know, chasing that kind of work because she knows, she does such a good job of that stuff. Okay, I want to talk about something that's hot off the presses. I've never, I haven't shared this with anyone yet. This, I think will be out by the time this episode comes out. I'm doing a state of the tech job market report. I do this every, a couple times a year. I've done it maybe three times now, where we basically look at hiring trends across tech. And we look at PM, engineering, design, things like that. One of the most interesting takeaways right now that I hope is out there by the time this is out is that PMs seem to be benefiting most in the AI era. We are seeing the most open PM roles of any role. Engineers actually slowing down, open end roles. Design is very flat, not looking great. User research, really not great. There's a hundred open roles for user research globally at tech companies right now, a hundred, eight, 108. So interestingly, when AI became a thing, many people were like, PMs are over. We don't need them. Why do we need them? Get rid of them. Surprisingly, it's the role that seems to be most thriving right now. And I actually had a post about this that I predicted this. I feel really good about this. So let me ask you this. Why do you think this is the case? Why do you think PMs are doing so well in the AI era? I think for a while, Lenny, if I could use a misconception I've been trying to debunk for a while with, especially even our customers and inside Atlassian, I wrote a post about it. But I think for a while, especially a year ago, it felt like the narrative was everyone's becoming an AI builder. And if the definition of an AI builder here to make it more useful is someone who spends most of their time building software with AI and AI builders is all you need and the roles are merging into one, et cetera. If I could share a case study, I think that adds to one of your data points inside Atlassian. So what we did was we're like, okay, maybe this is where it's headed. Like, let's look at it as a craft. Let's try to create a few squads that have AI builders. And we moved some of our product managers and designers to a team, a few teams that ran these experiments, basically build product where everyone was the AI builder. There was no PM or designer, but they came from different crafts. I remember checking in with one of the PMs and one of the teams and one of our one-on-ones in the first few weeks. And he's like, it's unbelievable, man. Like you're constantly building. It's so good. Like, I feel like we've moved so fast. We'll move faster than ever before. You know, early days, it was just amazing. Yeah, everyone's gonna be an AI builder. I'm like, wow, really? Like, is that where it's gonna hit? After about a month, he comes back and he's like, everyone's gone back to doing their traditional jobs. Like, sorry, what do you mean by that? He's like, it sounds so obvious in hindsight. It's like, we realized throughput's not our problem. Direction's our problem. Throughput's getting cheaper. So, you know, I wrote this blog post internally that was, you know, direct the ship, don't row it. And it was a lot of PMs coming to us saying, should I be coding? Do I need to code? And, you know, quite a few of our PMs have committed code in some capacity or whatever, but it's very obvious in a world where we have, again, the nuance here is if you're a startup and everyone's moving really fast, everyone needs to do anything, which I'm fully bought into that, or a zero to one project. In a company where you have the luxury of hundreds, thousands, tens of thousands of engineers, dozens of engineers, your problem is your direction, not how quick you can pump things out. So people who are actually helping everyone steer the ship, not more throughput, right? So I think that probably is in support of that, Lenny. I think more people are using this, more throughput, so I need more people to steer the ship. It's probably reflective of some of the results you're seeing on more PMs being hired. That is an awesome answer. And that makes, like, I imagine so many people hearing this are just nodding along. Yes, this is exactly right, in spite of what we keep hearing, just PM should be shipping code, PM should be building. There's a few directions I want to get here, I guess, just like the trend kind of to double down what you're saying, what AI is enabling is anyone can build, which is creating a ton of chaos and a lot of prototypes and a lot of ideas. And what you're seeing here, what this data is probably showing is just now the need is, what the hell are we actually doing? Let's all get on the same page. It's like more buy-in, we need more buy-in, we need more alignment, we need all the things PMs are great at. Also, what is great? Is this ready? Is this good? Is this what we want? Is this going to hit our goals? So it's like, now that we think about it, I see, this makes all the sense. Here's why we need more PMs. It's so interesting that we thought the opposite would happen. And we thought the same thing would happen with other crafts, Mert. I mean, Lenny, you were probably around, I'm guessing, I'm going to guess your age, Lenny, so no offense. You're probably around during the Balsamiq mock-up era, right? Absolutely. So like, what did we, the promise there was everyone can do mock-ups. Mock-ups are so easy, right? This is ignoring AI prototyping. And at last, yeah, Balsamiq exploited and we used a whole bunch of mock-up tools and it was, we liberate mock-ups. I learned so much through that era at Atlassian. I learned that one, yeah, everyone can create a prototype or a mock-up. But yeah, actually, PMs are some of the worst solution designers in the world. Like as in, you realize like, because you have access to the tool, doesn't mean you're an expert at the domain that you have access to, right? So like, now some PMs, more of the artist ones ended up doing just generally better mock-ups than others. So AI is really good at amplifying your strengths and equally it amplifies your weaknesses. So we learned there in that world that specialists is still important and specialization is still important. Like you can only get the AI to do something as good as your knowledge of the specialization is, if that makes sense. Cause you gotta learn how to prompt it. We started with the mock-ups era and now we're seeing it again with the prototypes era and everyone can generate a prototype. It's awesome. I mean, it certainly helps up speed up decision-making and all that stuff, but it's a fly in the wall with like all PMs and our designers. Yes, everyone's circles are getting bigger. Like think of it a Venn diagram, but the circles aren't overlapping. It's just, they're all equally getting bigger. So if they're equally getting bigger, you know, then they're not compressing, if that makes sense. The specializations are still needed to improve the quality and taste that's coming out of each of those things. Yeah, that's just often missed as well. I think that whole topic. I love this. Okay, so again, this is a fun spectrum. On one end of the spectrum of what a PM should be doing you are now an engineer, you're shipping production. This is part of your job now. The other end is don't even touch AI at all. Just do what you used to do. Like you actually spent a lot of time thinking about career ladders and calibration and what a great PM should, what a successful PM looks like. So this is something you spend a lot of time on. What do you think a PM should be doing in terms of AI and fusing their job prototyping? I imagine, should they be shipping code? What do you think is the right answer for PMs these days? I'll tie it back up to the top Lenny, which is their job is to make sure the team, they don't own decision-making, PMs own decision-making velocity. And so they're thinking of how can I help the team move faster? The second question is, oh, can I help in any of those? I think that's the right mindset. So it's like, if I realize that, okay, customer empathy and customer awareness is a huge problem, a context gap my team has, and I just feel like they're not making the right micro decisions because they don't understand this persona over that persona, et cetera, then you'll probably. be thinking about how can I involve them in the process, but also how can I use AI in the process to accelerate and filling that context gap. I think I would go back to the top and just answer that question of like, look at your team today. There's some things that you might feel like is gelling well and go, what is the context gap? And then look at how AI can help you do that. Is it because you don't have direction in the vision? So sure, prototyping might help you get a better vision and clarifying it and all that kind of stuff. But you might realize that they don't have a good understanding of the market dynamics and you might send off a bunch of agents, do some analysis and build that with a team, et cetera. So first stating the problem with the context gap and then using AI to help solve that I think is the right approach. I love that answer. Very unlikely it will be the bottleneck for decision-making and shipping code to production. Most likely that is the last thing that you need to speed up because that is so easy now. Unless as you said, if it's like a startup or just like a new team, or it's just a PM, for example, working on something. Okay, amazing. So while we're on this topic of AI, you hire a lot of PMs, you interview a lot of PMs. I love asking this question. What's kind of trending up in what you are looking for when you are hiring PMs? What are kind of emerging newer skills, habits, traits, and what's trending down? What's less important now? Trending up in certainly larger organizations, I would say, again, to be more nuanced in the topic here, is PMs that I'd say that are certainly more of the artist type. And why that's the case is because taste has become more important than ever before when everyone has access to the same intelligence. I think that's been covered heavily in your podcast. But I think it's certainly people are more leaning towards experience-led PMs. I think that's certainly trending up. The other big thing that's trending up, obviously, is if anyone is in an established workforce, as in they have teams that already exist, you're asking yourself, how can I reinvent how each member of my team does their job in the world of AI? The problem you have is probably cultural change. If you define culture, it's how work gets done. One thing I often do with the interviews and the interview process is you're looking at what new culture would this person bring to the team? And that means, how can they transform how we work, especially in the world of AI? So asking people how they work with AI is a fascinating question and getting to the deep beyond the, I use it to summarize stuff and whatever, and actually getting to that. Because you're also looking at cultural ads, AI natives, or how AI natives is such a relative term, because by the time you kind of, but you're always looking at what new ways of working can this person bring. So that's probably the second thing. And the third thing, Lenny, I think this is counter-cultural to some of the narrative that's out there, is around hiring people that don't know any better except for AI. And this may sound a bit spicy. Maybe their AI slop levels are higher, but they're actually, they didn't grow up with knowing anything else. In particular, I'm talking about juniors here. So one misconception I'm often debunking with a lot of established organizations is you don't need to hire juniors anymore. And I go, why do you say that? And it's because, oh, well, AI can do that for me, yada, yada, yada. But I'll say, but as an organization, what are some of the problems you're having? And often the problem is, well, I have a workforce that needs to transform and learn new different ways of working. And I don't know about you, but it's really hard for me to unlearn the ways I used to work and then learn the new ways of working. That's very hard. It's much easier to grab someone out of college that's basically been cheating in their university assignments or not knowing any better. Or I love using the example of my son, 12-year-old son. He vibe codes his own games. He's gone from scratch, like a visual coding language, straight to vibe coding. He's missed the whole syntax and whatever. He thinks that's normal. I'm always blown away when he'll dismiss a model and be like, dad, this Gemini wasn't that good. A Bobo wasn't that good. It did this. I'm like, dude, that would have taken like four days for someone to build. He's like, yeah, whatever. And I'm like, I'm blown away. And he's just like, you know. But the long way of saying is I think the third type of person you're after is, and I think in particular juniors for established workforces, they bring different ways of working to the team. Now, juniors are not great at identifying AI slop because they don't have the domain expertise. But your established team, your existing team, your seniors, your people that are specialized, they can pick AI slop a mile away because they know the domain really well. And so you're actually looking at how can I build a complementary team here, a complementary workforce between the juniors and the seniors so they can work together. So I would say that's the third big thing is we have definitely over-indexed on hiring more juniors than seniors because, again, our context is we have an established workforce, an established team that are learning to transform how they work, which is every organization in the world. And so you're going, what person can I bring in that doesn't know any better? And so we run these AI Builder Weeks, and they're fascinating weeks, Lenny, because we just take a week where all the crafts stop and they tinker with AI for a week, engineering, design, et cetera, at the same time is very important. The juniors or the younger people tend to be showing all this coolness and demos and tools, whatever, and the seniors tend to focus on identifying slop, evals, quality problems, et cetera. You just notice that just natural gravity, and you're like, wow, this is diverse teams, and diverse teams are still very important in the world of AI. This season's supporting sponsor is DX. Now that everybody at your company is finally using AI, the question that engineering leaders face is no longer, are we using AI, but is it paying off? With DX, you can measure and benchmark the impact of AI adoption across your full product lifecycle. DX shows you where AI is helping developers and where it's creating new bottlenecks. You can also use it to evaluate AI vendors, monitor costs and licenses, and see exactly what's limiting your agents. That's why hundreds of enterprises, including Snowflake, Sony, and BNY, use the DX platform to measure AI's impact on developer productivity. Visit getdx.com slash Lenny to get a demo of DX. That's getdx.com slash Lenny. Sharif, I'm loving this conversation. There's so much good stuff here. On this junior piece, very inspiring, I imagine, for folks listening to this that are earlier in their career or trying to break into PM from something else after school. Are there any kind of trends of where the best juniors slash young folks come from college-wise? Are they still going to college? Are they studying computer science? Are they MBAs? Anything there? I don't think I have anything unique to share here. I think it's still a mix of computer science. I went to our APM cohorts last week in Sydney, and we had them all together. There was some people with arts degrees, someone with a law and technology background. It was a good mix. Again, you're looking for the critical thinking, the usual stuff. The curiosity is super high, willingness to try different things. They've got no fear. They're just willing to try anything to try to make it work because they haven't done anything any other way. We have numbers to show this, Lenny, both internally and externally with our customers, our research team. Something like almost 30, I think it's 38 percent. I posted a blog about it, but juniors are more likely to adopt new tools than the seniors, both internally and with customers, et cetera. On the flip side, the seniors are getting more quality and better productivity out of the ones they're using, but they're less likely to try different ways of working. If I were to connect, you have three answers of what is trending up. Artist PMs, people that bring a culture of open mindedness and trying stuff and experimentation because things are changing so fast, which feels actually pretty close to the third bucket, which is newer workforce folks that are just AI native, essentially. The Venn diagram is essentially artist PMs who are new, juniors, let's say. Going back to the artist PM, say a little bit more about what that looks like, because people listening to this are going to be wondering, am I that? How do I become that? Talk more about, as one, as you self-describe yourself as, what does that look like? What do you grade at as an artist PM? An artist PM is, in contrast to a general manager or a scientist type of PM, the way they lead product discovery is largely through visions. They may often actually start with solutions and work back into problems, which sounds, again, very counter-cultural, because they have some intuition in some particular area. They tend to focus on experience more than numerical. They're more data-informed than data-driven. I want someone in growth to be a little bit more data-driven than data-informed. You want that healthy tension, right? They'll tend to do that a lot more. A good signal is you'll see a bunch of prototypes from different PMs. The artist ones will generally do the ones that seem more logical or flow better. You'll find sometimes, I always say, when you're in a room and you're discussing a concept or a topic and someone sometimes feels the urge to pick up the wipe or marker and draw, that's a pretty good signal. They're like, are we talking about this? Are we talking about that? They'll often tend to do that. But I think in the world where more vendors and more software is cheaper to build, more people will have the same sets of features, experience matters more than ever before. I'm not a believer that everyone's going to chat to an agent and the agent interface is the final interface. I think there'll be many. I think experience still matters a lot. I think that in a world where just generally speaking, more people can build more stuff cheaper, faster, easier, experience matters more because that becomes one of your possible differentiators. Those artists tend to be a better asset, but not a better, a good asset in that world. And so you're trying to complement your workforce to make sure you've got a mix of these skills. I imagine it's hard to actually find that combination of a new entrant into PM and very good at this because my suspicion is you need to do this for a while to build that gut feeling and the artist experience. True or not true? Someone just sent me, we asked all our APMs from last cohort, the new cohort coming up to record a loom of a take-home exercise. And I watched a few of them a week before last, actually. And it was fascinating, again, to see how, again, they don't have domain experience, but we gave them an exercise to build an app for a particular thing. And to see how they chose to present it and lead the narrative of why they're building this particular app, whatever, you do pick up the artist streak in them very early on. You can see the ones that will, as an example, there was a loom of someone that went through and they walked through basically graphs before they even got to any experience. And they were like, here's the business case, he's the TAM, he's et cetera, whatever, whatever, whatever. And they were like his experience. And they spent very little time on that in contrast to this other candidate. And she walked through the end-to-end journey and talked about, it's funny, she talked about delighters and like, I think this would be a way to build an emotional connection with someone. Again, don't have any domain experience, but you can see the way they tended to do product discovery tends to be more visual than the other person. So you can at least identify that streak early. I mean, Lenny, I've got two kids, my son's really good at maths and logic and whatever, and my daughter just loves drawing. And just naturally, we brought them in the same household, we read to them the same books as kids, at least I think we did. But you can just see the natural, their preference in their style of how they work and how they think. And so I think that there is a bit of nature and nurture here as well. Yeah. So the implication here is, it's not like you train to become an artist PM archetype, it's mostly that is how you are as a person, and then you build enough experience to actually be good at it. I know, I'm sure. Tell me if you disagree, but it's probably like, this is you, and then you can become better at it, and you can probably learn how to do some other things better. But I don't know. Yeah, you're nodding along. Yeah, I think it's easier to accelerate the thing you're naturally good at, and it's harder to accelerate the thing you have to learn to do. I think that's applied to anything. So is it possible to train artists? Yeah, totally. And you can at least help people identify things or whatever, and it's just harder. It's not as easy for that to do. I love how it's another counter-narrative. I think that's the word you've been using of like, okay, we don't need junior people anymore where they're done, there are no careers left. You could argue they are the only people that we'll want to hire now. I don't think that's what you're saying, but it's like, these are the people that are so open-minded and so native with what is possible now and not stuck in this old way, that that's maybe who we need. And the people stuck in the old ways are only slowing us down. It's the theory of diverse teams is still true today, and even more true in the world of AI. Like, it's just so true. So this data that I've been seeing of designer roles not growing, there's just been like a stasis in the number of designers being hired in tech. I wonder if it's because basically PMs are filling in more of that gap, as you've been describing, PMs basically kind of being the... And I'm not saying designers are not very useful and amazing and important. And I'm surprised that designers aren't thriving more in this era where, as you said, the experience is a huge opportunity and differentiator, but that's just what the data is showing right now. And I wonder if it's because PMs are filling in more of that gap. I'd love to dig into that one. I think there's some other hypotheses I would have. Another hypothesis I would have, I don't know if this is true or not, is that the market still hasn't got past the aha moment for experience matters more than anything else in the world of AI. Maybe if you rewind back, we said all PMs are going to be builders one day, and I think we're learning that's certainly not the case. I think there's still a misconception to be debunked or proven that design is more important in the world of AI than ever before, because if everyone has... is the same thing. One of the only things that will matter is why you prefer something. I think that's just a reflection of possibly where we are. That would be my leading hypothesis. I can just say at Atlassian, design has never played a more important role in roadmap driving than ever before. I've been here for 17 years and I am fully convicted by it. Experiences will matter most. I love that. Let's takeaway. Stock up on your designers right now. Buy low while no one's realizing that's the opportunity. Slow it up. Put them in the garage with your spare water. I've been feeling the same thing and it's interesting. Your feeling is maybe we're not realizing that this is a huge opportunity and a way to differentiate and win. Yeah. We'll see if that theory is correct. That's my current hypothesis. Cool. Okay. I'm going to go back to a couple of these counter narratives you mentioned, because I think they're really interesting. One that you touched on a bit that I think is worth spending a little more time on, which is not starting with a solution. PMs are constantly told, do not start with the solution in mind. Start with the problem, with the customer pain. Don't even talk about the solution yet. All that stuff. Your feeling is that's not actually true in a lot of cases. Talk about what you've experienced, what you've learned there. Yeah. I think it's, again, one of those misconceptions. It's a classic easy thing when you're in a meeting and someone says something and if you want to look smart, you'd be like, well, slow down, everyone. What's the problem we're trying to solve? And that's like, yo, that person really brought something. Take a step back. Yeah. Nice, Lenny. You got that one. It's a good question to ask, honestly. Yeah. What's the problem? I think it's still a good question to ask. I just think what's not talked about enough is sometimes it's a great way to start with a solution. A few examples, if I could run through them, I think one, sometimes it's okay to start the solution when you're trying to discover, you feel like there's a hunch in the area, you're trying to discover what the things you could productize are. Maybe some recent examples, at Atlassian, we have RovoChat, which is our AI assistant if you're using our platform, and RovoAgents. They're agents you can deploy in loops inside Jira workflows, Confluence workflows, and so forth, et cetera. And believe it or not, Lenny, again, three and a half years ago when we created agents in RovoChat, three and a half, yeah, geez, AI, it's gone fast. We were debating if we should put these in there because the theory was, well, why would we put general chat in there and why would we put agents in there and let people create their own agents? Why don't we productize, build the feature that people want, then give them the general canvas? So let's not put chat in there. Let's put a summarize button on top of Confluence pages rather than having a chat button to do that. And what you often learn is, actually, if you give people a canvas, that's sometimes the fastest way to discover what the main problems are and productize some of those problems. So as a simple example, in Confluence whiteboards, customers can draw sticky notes and collaborate with them, whatever, and one of the common scenarios is you might use RovoChat to group sticky notes into common themes. Now, chat's not, maybe to tie it back to our experience discussion, chat's a terrible experience to do that visually. You're typing, you're filling out, then which sticky notes do you want to group into themes? How do you select them? And so, obviously, you can productize that feature and that team lets you select a bunch of sticky notes and decide, I'm going to group these ones into different themes, and AI goes and does that for you. A really good example where you productize the use case that chat had, but chat was the infinite use case. Another example is customers love putting, you have nearly 10 million agents that run in loops every month inside Jira and Confluence, and you want to give a ticket to an agent, and the agent picks it up and does the triage, or agent generates the code, or agent refused the legal contract, whatever it is. That's quite common inside Jira, people put agents in loops. And so, we can learn what kinds of agents they're putting in loops, and for some of them, we're going to leave them as the infinite use case, let customers build their own. And for some of them, we're like, actually, we're going to build an AI triage feature and make that native to the product. And so, starting with the customer solution and understanding what the customer is doing helps you work out what the problems are, and then you might productize their manual workaround into a set of capabilities into your product. It's a pretty common tactic. I think I've heard some people refer to it as paving the cow path. There's a walkway and there's a cow that walks straight in the middle across the grass, and you're like, well, why don't we just make that the cow path? And so, starting with a customer solution, there is a really good approach. The second big one really is strategy, product strategy. The best product strategies are starting by looking at what assets or what solutions do you already have today? Because you need to answer, where do we play and how do we win? And to answer how do we win, you have to look at what unique things that we have as a company, as a business, as a product, right? So, the example I love to give people is the Amazon bookstore is such a good example, who built this ridiculously sized bookstore. It's a huge inventory. If you read Jeff Bezos' book, he's like, what other things, he's starting with a solution, what other things need a ridiculously large inventory that is a crazy scale? Or what if it was an everything store? Do you know what I mean? So, starting with a solution for product strategy thinking is another good tactic. I always say to teams, write down what assets you have today, then write down your beliefs and see if there's any connection between those two of where things will head and do that. So, giving people a canvas to try things and productizing the solutions that they manually built is a good strategy. And so is when you are doing product strategy, start with what solutions do you have today and what unique problems can you solve with them? Got it. Okay. So, it's nice to have, here's the paths where starting with solutions is good. You're not saying always start with solution or it's always beneficial. And essentially, in most cases, you actually don't, except in these two situations. Yes. Actually, I don't know if it's most cases, but it is important to understand the cases. I would say that a lot of customer discovery is by learning what customers are trying to do with your product and then trying to make that better. So, you're actually observing a customer trying to create a solution for themselves in most products. And then you're like, oh, let me make that better for you, et cetera. So, it'll be fascinating to try to come up with the percentages that most of the time, you're actually implicitly starting with a manual customer solution, if you like. I think that's a good thing to think about as you go into do exploratory, which by the way, with AI, it seems to do enough of just talking to humans. It's become much easier to stay behind more dashboards. By the way, I didn't know you have this product, the whiteboard product. I didn't know that was a thing. So, this is like NeuroFigJam-ish? Yeah. Confluence whiteboards. It's huge. Yeah. Customers can embed whiteboards in Confluence and they add agents to them and they collaborate, real-time collaboration, add sticky notes, turn them to Jira tickets, all that kind of stuff. So, this is like a feature within Confluence. Okay, cool. All right. Folks don't know that it exists. I didn't know it existed. This is great. Good marketing for the whiteboard product. Didn't intend it to be as marketing, but it's a good example, I think. Yeah. Okay, cool. There's another counter narrative that you have, which is around AI, and this kind of concern people have with just bolting AI onto stuff, and your feeling is that's actually good in many cases. Talk about that. This is where the catchy headlines, Lenny. It is very catchy to say, don't bolt AI into your product. You should reimagine the workflow. I certainly believe that. However, the nuance is, I think if you have an established product, you have something ready today, the best way to reinvent your product to be AI-native is to understand the flows that exist in your product and literally bolt on AI to them, which sounds dirty. You may not market it that way, but actually, just think about it for a second. You've got a product, the product has millions of users. What I see some people doing, and I talk to a few startups or a few product companies, they're like, we're going to rewrite our product from scratch. I'm like, yeah, I think you can do that, but isn't the better way to actually work out is how to evolve, not trolly transform? The example I love to give is the Rover Agents one I just told you about is a good one, which is customers have already modeled billions of business workflows in Jira, just as an example. Jira is largely used by many non-software teams. It's a misconception most people have. For a while, you get an app approved in your app store, went through a ticketing system like Jira, it was Jira for a while, and then it goes through a whole review process. If you complain about your particular service with a telecommunications company, it goes through Jira ticket. Customers have billions of workflows already modeled on our platform, as an example. When it came to understanding how agents help humans move work forward, we looked at what are the human workflows that exist in the product? That's a good formula to go, what are the human workflows that exist in my product? Some of them are rigid workflows, as in someone's literally defined something. Some of them are informal workflows like lowercase W workflows, they're humans doing something. Then you can go, where can agents assist with steps of the workflow to generate new value, improve efficiency or something, whatever it is. In Jira workflows, we basically added the ability to add agents to columns and cards and whatever. Is that bolting on, Lenny? Let's use another term, maybe. If I was building an AI-native work tracking tool, that's probably what I would do anyway. I would build a workflow and then let agents work on bits of the workflow. My message to teams out there is like, hey, yes, reimagine the workflows. You should do that as an exercise, but look at where your customers are today. If you have an established user base and established users, your job is to help them move from AI-novice to AI-native. You're trying to transition your customer cohort across as well. Think about where they're working and where you can slowly introduce AI in those workflows. Again, another misconception there, I think that from a marketing perspective, yeah, you probably wouldn't say bolt on, but we literally let you bolt on an agent to a Jira workflow and you stick it in a loop. Then agents work in loops and they run millions of times and whatever. They're measurable value. If it doesn't work, you kick the agent out or you make the evals better for the agent. There's a whole bunch of stuff you can do there. I love all these takes. Before we started recording this, I asked you what's the goal you have for this conversation? You talked about how we need more nuance in the conversation of building product. There's all these hot takes, these viral clips that go out and people are like, oh, that's how you have to do it. I'm guilty of people on this podcast saying a thing and then people at a company taking that and this is the way to do it. Lenny's podcast said that's the way to do it, which I hate. I love that that's what you're trying to do here is add a lot more nuance to what people think is true and show that it's not. We've got all these spectrums and gradients that we've been talking about. Maybe just say more about that, why you think that's important. The classic thing in product, I think especially, is every advice you've probably got on how to do product, someone probably has a counter, totally opposite advice that works just as good. Again, you as a product team or a PM or a product leader, you're trying to understand the context in which their advice worked in. That's where a lot of the nuance gets missed. The AI builders one, Lenny, that we just talked about is a great example where I think zero to one projects, even in large organizations, you should put an AI builder type onto it. I'm fully convinced of that. But again, most organizations that have access to developers, this is not their problem. Their problem is where to direct the ship. The context in which the advice is given is super important. Just start with the solution, not the problem. I would equally advocate to start with the problem, but not enough people talk about the counter thing. It also just gives people confidence in like, am I doing things the right way? And I'm like, is there a right way? If you do something and you've learned something, share it with the world. For a while, I used to try to write blog posts about, I would share every craft lesson I've learned. One of the most things we're talking about internally is how do we scale product management inside Atlassian? I go back to when I started Atlassian, there were no product managers on Confluence. The way we did that was these things called product engineers. I learned what a product engineer was, and I had realized there was no term codified out there. I wrote a blog post about what I learned about product engineers, whatever. You realize product engineers help product managers scale. They help you. They're the champions of that shared context that we talked about earlier. All those nuances, I don't think get talked about enough, but I appreciate you running these talks and hearing people out share their nuances. I think that's what needs to happen more. It's tough. I'm in a tough spot sometimes where people share an opinion and many people are like, that is wrong. Why did you not push back on that? I find the best thing I can do that scales is have different perspectives here and not try to get everyone in line. This is the way to do it. People can decide, here's what fits for me. Someone actually built a really fun project. I don't know if people know this. If you go to lennysdata.com, you can actually connect to an MCP that has all of my podcast transcripts. You're going to be in there. You're going to be slurped into this MCP soon. It's also all my newsletter posts that you can use in any way you want. Somebody built a tool that shows both sides of every debate. Oh, that's awesome. Like the contradictory advice? Exactly. That's so good. I'd love to see that dashboard and you'd love to see why it worked. This advice and why it worked and this advice and why it worked. That's awesome. That's so good to see. Is there anything you want to say that'll feed into this MCP that'll be forever part of the dataset? I think I've given quite a few misconceptions here. We've talked about misconceptions of AI builders. Everyone's a builder. We talked about the misconception of all PMs are the same. I think artist, scientist, general manager is one framework. You could do that. Misconception of PMs make all the decisions, right? I think that's another one we talked about. Oh, management. Management's another one, Lenny. So everyone runs into their career in product management, and usually in most organizations, you hit a path where you, hang on a second, I want to be involved in the strategy discussion, the strategy offsite. Oh, but it looks like the people that go to that thing or do that activity seem to be managers. I went through this experience myself. So therefore, I need to manage. And they hit this fork in the career where they feel like that's the decision. And so the other, I think, misconception when you help the bunkies, I think we try to do more of this at Atlassian. We have multiple distinguished product managers and engineers now where they work at the executive level. And so we're huge champions of the dual track career paths for individuals and managers. And a lot of people switch between both depending on the season they're in. Oh, they're starting at zero one. They'll go to managing a small team for a while, and they'll get big and they'll move back off. And we should celebrate that more because it plays to that person's strength. And so that's another misconception. Like to be involved in large strategic decisions at exec level at large companies, you need to be a manager. Now, we have a lot of individual contributors that contribute at the exec level. Well, sure. A counter narrative that I've been feeling, which is that there's always talk of we don't need managers and that managers should be ICing as well. And I feel like that's just not a good idea. I feel like managers are actually very important. And I feel like to be a good manager, it's very hard to also do meaningful IC work. It's really hard. And so you can take two worldviews there that you could take. If you're going to manage, manage a lot. So take like 10, 15 people. Or we want managers that will take three or four and own a project as well. And the reality is we have a mix of both at Alessian. And again, try to play to each person's strength. And that's how you'll get the best outcome. And so do we have the span of control for managers? We generally try to say you should have at least three or four people minimum. I think most are in the eight or nine. I could, something like that. But yeah, you do have some managers that will take on a few people and will own projects. And some that will take on a lot and play a coach role a lot more. Let me ask you this. For new PM managers, so you just got into it, what do you find is the biggest issue that they run into slash thing they should probably be working on? The number one thing I've seen is they go too hands-off too quickly. And they think that's the best behavior. It sounds a bit counter-cultural. I think the best leaders that we've ever had at Alessian that continue to be there, the best managers, I shouldn't say managers, managers, leaders could be two separate things. You could have a leader who's a good manager, a manager who's not a good leader and vice versa. So the best managers who are great leaders are able to go in with the team and help them get through outcomes. I can sit in a product review, Lenny, I'd be like, this isn't simple enough. And it's not like the team was sitting around going, all right, let's, sorry, I forgot to write simple to the agent. But the best managers are like, let's explore some options together. And they get in and they help the team work together and get through the outcome. I think that would be my number one observation. They got so hands-off and they're like, I'll almost zoom out completely and let the team do that, which is good. You need to create space. Don't get me wrong. But the teams love it when they see managers lead by example. So I think it's going to be more important all the way on. That's great advice. Okay. Possibly final question. I've heard from a bunch of people that you work very closely with the founders and execs on the big product keynotes that the founders do. I love keynotes. Something most people don't say. You also work with the CEO and co-founders on their founder presentations, things like that. What's something you learned about working with the top of the top that might be helpful for folks that maybe struggle working with execs? There are so many. Dangerous last question, Lenny. Without a doubt, I describe working with someone who's a strong visionary leader, often in founder-led companies, that's still the case, et cetera. It's like me and you watching a movie and they're on scene, chapter 10 of the movie, and you're on chapter three, and the rest of the teams are on chapter one. And they're saying stuff or whatever, and you're thinking, wait, what? Are you sure? That's everyone's first reaction. Now, you can do two things at that point as someone who works with that stakeholder or leader or vision leader, et cetera. You can say, I'm going to follow orders and go do that and just pass that on. But the dangerous thing you do that is you're doing that with no conviction. You're doing that because the boss asked me to. Or your job could be like, by the way, the vast majority of the time, you could argue was the stakeholder or the founder right or wrong. In my experience, they've been right 99% of the time. It's just what I've realized is they've thought about that thing far more than anyone else. And the aha moments that got them there, my job now is to fill that context gap to work out why did they get there? And do I feel convicted of their thing? And I'll often start by like, I don't know, man, I'm still working through this. What made you leave that conclusion? And I'm trying to get to chapter 10 as quickly as possible. And once I get to that, I know now what I need to do to bring the rest of the people along with me. So I would say often with these leaders, they're thinking about things a lot and start by trusting, but it's not. I think the counter narrative here is the HIPPO, Highest Paid Stakeholder, whatever, I can't remember the acronym, is the highest paid opinion. Highest paid person opinion. Yeah, I can't spell HIPPO. I totally butchered that one. Highest paid person. Yeah. So I think the flip side is, oh, there's some stakeholder who thinks this is their pet project and whatever. I think that's certainly true in companies. Don't get me wrong. But I think when you have someone that you know that has thought about this domain particularly well, especially in the enterprise B2B space, I always say to someone, if someone's worked in this space for several years, it is vastly different to I did a year here and I did a year there. It takes time to make large changes and whatever. And so assume they're right and assume you're wrong and try to fill that context gap. And sometimes you'll prove them wrong, but it's not the goal here. The goal is to get the best outcome for customers. And so then you can bring the team along on the journey. Let me ask you what happens if you actually disagree? Because I think a lot of people feel this, go through this, and they're like, that is a bad idea. Tips for handling that? So if you feel like you've caught up to the aha moment and you're still disagreeing and you feel like you have a good case for why you disagree, I will still state that. And then I'll state, is this something you have crazy conviction for? And I'll say, are you the decision maker or am I? I'll ask that. Common question is, who owns this decision? Just be super transparent. And if they really feel like, I would like to take this decision or I want you to decide, I disagree, but I want you to decide, which by the way, they do a lot more than you'd imagine, like they'll be like, I disagree with this, but I want you to own that decision. But I will back you if I disagree and I will back you and move along with it. So transparency is key to foster that relationship. So, but what do you do when you're communicating it to your team and you disagree? Anything there? Is it stand behind it always? No, I'll present both arguments. I will state why I believe one argument is more superior than another and I'll equally argue for the opposite. There was a recent example where we're branding a particular thing and whatever, and I think it was like the best of the worst decisions, best of the worst options we could make. And I'll say, I think here are the pros, here are the cons. And if you argue for the opposite, I think here are the pros and here are the cons. We have decided to make this decision. I'll try not to call the individual out. And they're both legitimate trade-offs. And we're moving ahead with this one. And you need to speak the room of what the other opposite argument that everyone's thinking and represent that appropriately. The best PMs I find can counter, like argue against themselves and present you really good reasons why they shouldn't do the thing. The really blind ones will be like, oh yeah, my thing's the best. It's, you know, we should definitely do this option. It's the best option in the world. They're like, I often say, why would we not do this? I love asking that in reviews. And then sometimes like, I haven't thought about why we wouldn't do this. I'm like, you should argue for the opposite as well and feel like you've intellectually wrestled with that and then move forward. That is great advice. Something that I'll mention briefly, when I found, when someone with power shares an opinion, a lot of times people take it as that is what we need to do. And Dharmesh Shah has this really great framework. He calls flash tags. I don't know if you know this, where he basically says, okay, he's like hashtag FYI. So when he shares an opinion, it's like, okay, this is just an FYI, or this is a suggestion, or this is a recommendation, or this is a plea. And it's a nice way for someone that implicitly has power to communicate, okay, this is how I feel. Just an FYI. You don't have to do this, or it's a plea. Please do it. Yeah. And I'll say, those are like our frienders do that as well quite a bit. They'll say strong opinion, not my decision, blah, blah, blah. I will tell, but often you'll get people just in a discussion, they won't pre-hashtag their discussion as they talk. And so I'll say to PMs, your job as a PM is to be that person in the room to ask that clarifying question on level of conviction on the topic. And just to be clear, the classic playback technique at the very end, but as part of that process, be like, this is a conviction we will go and explore further, or this is something we will take as input and then medicine ourselves and be clear with that. So I'll put that back on the PM sometimes because the stakeholder doesn't always do the hashtag at the end of, or before, or whatever it is, if you know what I mean. Oh, man. I feel like we could talk for hours of all the tips and tricks and tactics you've learned. This could go on forever. I'm going to end it here so that we have time for another episode someday to go deeper. But before we get to a very exciting lightning round, is there anything else that you wanted to share? Anything else you want to double down on before we get to our lightning round? I think to your listeners, nuance, I think we've talked about the nuance is important. So please share your stories. I would think the other thing I would just highlight is you're a product of someone else creating opportunities for you. So ask yourself, who are you creating opportunities for? And if you can't think of that name listening to this podcast or watching it, try to find someone that you can champion, be a voice for, et cetera, because you believe in them and you want to give them a go and that kind of stuff. So just remember where you came from is a long way of saying that. I love that. Well, Sharif, with that, we have reached our very exciting lightning round. I've got five questions for you. Are you ready? Let's go. Let's go. First question, what are two or three books that you find yourself recommending most to other people? Look, I read with my kids a lot or read books with my kids. Lately, we're reading The Prison Healer, Yvette Noni, I think her name is. It's like a fantasy book in a prison scene, dystopian prison kind of escape story. Awesome. Read with my kids. Like I personally read the Bible a lot. I think it's a book that I just keep coming back and meditating to and gives me some kind of grounding in the world of where AI is headed as well. But most of my books are kids' books. The other one I read with my kids, Winged Feather Saga, is such a good read. It's about a bunch of kids who are heirs to a kingdom and they don't know that and they overcome adversity to get there and stuff. So I read a lot of kids' books. Favorite recent movie or TV show? Project Hail Mary was awesome. I mean, I cry in a lot of movies, so it's probably better. Yeah, just totally loved it. Awesome watch. Very thought-provoking, emotional. Fun fact, my cousin is the Russian scientist in that movie. That's awesome. A little somewhat distant cousin, but still related. She's killing it. Okay, next question. Favorite slash most interesting AI product that you recently discovered or just are really hooked on? This is total recency bias. I mean, there's a few favorite AI products, but I'm just going to go with the massive recency bias. We're on a long drive with the kids. Google Maps added Gemini to Google Maps, at least in Australia, in Apple CarPlay. I don't know if you've seen that before or not. So when we drive, we typically put on, sometimes we put on music, sometimes we put on AI and play a game. We play a road trip game, or we'll play a quiz game, or we'll take turns, or whatever it is. Everyone will be yelling at the AI while we're driving. It's very entertaining. But it was fascinating to see in the map view, Gemini added, because it's such a lesson in product, workflow, proximity. I'm in the map, I'm not going to close it off and switch and open one of the AIs to talk to it. I'm just going to chat to Gemini right here. But it's super contextual to the trip. So it'd be like, hey, we're heading to this destination. And we started playing some fun quiz games about the destination we're going to, etc. And it was just one of those delightful experiences in the car that I'm like, that was awesome. It was a fun experience. I could still see the map and talk to the AI at the same time. The kids had a blast. So it was entertaining. Got them off the screens for at least 30 minutes. Did they just bolt AI onto it? Oh, nice. I like how you did that. Callback. They bolted it onto an existing workflow. Some of the best strategies you'll ever do. Nice. Nice, Lenny. I love that. Example of you being right. Here we go. Okay. Two final questions. Favorite life motto. Is there a motto you often come back to in work or in life that you find really useful? People tell me I'm very relaxed and fun at work and always joking. It's usually how I deal with my nerves, to be honest, Lenny. But they ask me why, etc. And I'm always reminded that you as a person, decouple yourself from your job. identity, which is really hard, especially in the world of tech, and celebrate the individual outside their work life, I think is an important thing that we don't do enough, and I love playing footy, which is like flag tag, with a bunch of folks who live locally and none of them work in tech, and they're just a mix of electricians and bricklayers and whatever and different jobs, but like they're amazing human beings that don't work in technology, and so reminding people to decouple themselves from their job as their identity I think is something that's healthy and will keep you sustainable in your work and in your personal life as well. I will share a quote that has come up a couple of times on this podcast that always hits people right in the heart, the only people that will remember that you worked late are your kids. So true. I've heard a simple one, which is nobody goes to a funeral and a tombstone, like it's saying a tombstone, this person worked so hard or whatever, like that's never in the eulogy or whatever, so that's another motivator. They hit the quarterly number. All right, final question, I've been wondering what is going on behind you? There's like all these fun things that are happening, what is this, explain. In Lassie and T-shirt culture, we make a T-shirt often to tell a story about a moment in our history. So this is like Confluence Questions was an app, that was when we had our first QA team, that was when we launched a Vietnam office, there's a story behind each one of these shirts. And I love collecting them, it's just a great way to tell stories, when the bigger you get, the stories become more important to tell the teams, and I've got hundreds of these shirts and one day my wife ripped up my shirts and surprised me for a thing she did for my birthday and created up this giant plaque, this huge shirt and stitched together some of my favorite shirts that have stories behind each one of them, so that's what that is. That's very sweet, I could see that going horribly wrong. What did you do to my shirts? I've been saving these for years. I wrote a blog internally that was titled My Wife Ripped Up My Lassie Shirts and people were like, what? And then they read the details and whatever, but then for our recent 20 year anniversary as a company, I just walked through 110 shirts and recorded a video on YouTube, it was a lot of fun. And just like the story behind each one, it was just very satisfying. Let me ask you this briefly, what's a unique cultural kind of ritual of Lassie that might be fun for people to learn of? There's a lot. We're an incredibly transparent company, open company is one of our values, as in every exec review is usually recorded, available to search and find and loom and ask questions and run the AI agent through it. So the transparency here, everyone that joins says, I've never experienced anything like it. I mean, I haven't started Lassie for a long time, so I'm used to it, but they're like, oh my gosh, when do you stop reading? There's so much content. So I would say open by default is the, ironically Lenny, I think it's also, I mean, partly because of confluence in the nature of we have of the product we have, people just document and create and it's open by default. But sharing context is key in the world of AI and so having an open by default approach to things is actually better in the long run for you as a company, because by definition it means there's more context that they are, has access to and the humans have access to. So the openness is, is unbelievable here, but it's, it's pretty crazy, but people love it. That is, that is super interesting. There's, there's been a number of companies over time that have done this, GitLab is a good example and it's interesting now with AI being able to read all of the things and tell you what you need to know, it changes that. And like in theory, you could like, it might expose things that maybe people shouldn't have seen much more easily and quickly. On the other hand, it gives you all the context as you described to make better decisions. I have this theory, Lenny, so we have, we have a team anywhere policy, so you can work anywhere you like. We have a legal entity. So some people choose to work remote a lot, some people don't. And so just as an example, you know, when I have part of my team is in the West coast of the USA, they, a lot of them like an in-office culture and for reasons, and there's some people in the APEC region love a distributed culture. I have a theory we're trying to prove out right now, hypothesis, which is those that are remote are more likely to get more outcomes out of AI than those that are co-located. And so we're just got a bunch of data scientists trying to prove that out right now. So that'll be a fascinating theory. And the thing here is that you just tend to write more and share more digitally when you're remote. And consume. Yeah. Yeah. And you know, you can do whatever. And they're both good approaches and you need both, but it's, that's a, again, maybe another counterpoint of the going back to in-office culture, maybe. Going back to this report that I'm putting out, actually, remote work has plateaued. There's no, it's kind of stalled at 10, 15, 20% across industries. So that'll be interesting to watch if that grows again with AI. Anyway, I keep asking you questions, but we should wrap up. Shreev, this was awesome. This was such a great, deep, important conversation that I think is going to help a lot of people. You're talking about opportunity. I think this is going to open up a lot of new opportunities for folks. So thank you so much for being here. Is there some way that listeners can be useful to you? That's my final question that I almost forgot to ask. If you see something in social media, a lovely tag headline on one line, grab, you shouldn't hire juniors anymore or whatever it is, whatever it is you see. Be the person that asks the nuanced questions, be like, oh, why did that work for you? Why didn't that work for you? Don't be the person that amplifies that one view without first understanding the context. I think that applies to many things in society, not just product, but yeah. Nuance. Yeah. Amazing. Shreev, thank you so much for being here. I appreciate the time and the opportunity. Thanks, Lenny. I appreciate you. Bye, everyone. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at LenniesPodcast.com. See you in the next episode.