Glean CPO on why shipping more features with AI doesn’t mean you’re creating more value

Emrecan Dogan is a trained industrial engineer who has spent his career applying systems thinking to product management, and as Chief Product Officer at Glean, he has found the perfect laboratory for it. Glean is an AI assistant and platform for the world’s largest enterprises, and the same principles Emrecan uses to think about building AI products, where fixing one bottleneck immediately moves the constraint somewhere else, are the principles he uses to think about how product organizations should be designed and how PM careers should evolve. In this episode of Product Talk host Sonjoy Ganguly sits down with Emrecan to talk about why the correlation between how many features you ship and the value you create may actually be getting weaker in the AI era, why he believes the super IC PM is the defining career opportunity of this moment, and how Glean runs a product org where autonomy is taken, not given, and where the product itself tracks organizational intelligence in real time.

Join us for new conversations with leading product executives every week. Roll through the highlights of this week’s podcast below.

Show Notes:

1. Emrecan trained as an industrial engineer, and that background shapes everything about how he thinks about product management. Industrial engineers study systems and bottlenecks: the principle that improving one part of a system does not improve the whole system if the constraint just moves somewhere else. He applies that lens directly to what AI is doing to the PM role, and it leads him to very different conclusions than most of the conversation about AI and product.
2. The popular story about AI in product is about coding speed: AI makes it faster and cheaper to write features, so teams can ship more. Emrecan’s response is that coding was never the only bottleneck. The moment you accelerate code production, the constraint moves downstream to QA, to documentation, to system architecture, to the question of how all these new features compose into something coherent for the user. And it moves upstream to roadmap: if you can ship faster, you need better insight into what to build next or you will run out of good ideas quickly.
3. His conclusion is the opposite of the pessimistic take on PM careers. He thinks this is the most exciting time in the history of the profession, specifically because the bottlenecks that AI is not solving are exactly the bottlenecks where great PMs create the most value. Customer empathy, judgment, systems thinking, and the ability to decide what not to build are harder to automate than spec writing, and they are becoming more important, not less.
4. The role Emrecan is most excited about is what he calls the super IC PM: a product manager who, rather than hiring more people to scale their impact, finds more and more ways to use AI to extend what they can do individually. In previous eras, a PM who wanted to take on more would build a team. Now some PMs are discovering that they can take on larger and larger surface areas while staying as individual contributors, with compensation that grows alongside their scope. He thinks this is the defining career path of the current era.
5. His favorite interview question for PM candidates is: what is changing significantly about being a PM right now, and what is staying exactly the same as it was five or ten years ago? He uses the question not because there is a correct answer, but because it forces candidates to demonstrate whether they can hold two things simultaneously: an accurate read of what AI is actually disrupting, and a clear-eyed view of what human judgment is still irreplaceable at.
6. The one thing Emrecan says has not changed and will not change is being customer-driven. He defines this as a particular combination of selflessness, curiosity, and relentlessness. Not just caring about the customer, but being genuinely inquisitive about what you do not yet know, and being relentless in the pursuit of uncovering new insight about what creates value and what is missing. That orientation is the same whether you are managing a single feature or leading a portfolio.
7. What has changed is that AI can now automate or accelerate large portions of the traditional PM workflow: voice of customer analysis, bug fix triage, technical documentation, PRDs, prototype generation. The question this creates for every PM is not whether to use those tools, but what to do with the capacity that gets freed up. The PMs who will matter are the ones who redeploy that capacity toward the things AI cannot do, not the ones who use it to ship more of the same.
8. On the CPO role specifically, Emrecan challenges a common framing. Many first-time heads of product come into founder-led companies believing their job is to define the vision. He thinks that is usually the wrong orientation. In most cases, the founder’s vision is already there, embedded in how they think, talk, and make decisions. The CPO’s job is to extract it, make it coherent, and translate it into a scalable narrative that hundreds of people across the organization can execute against every day without losing the thread. That is not a lesser job than defining vision. It is a harder one.
9. He also sees the continuous triangulation demand on PMs as one of the fastest-rising aspects of the role. The market, the competitive landscape, and customer needs are all changing so fast that a quarterly vision update is no longer sufficient. By the time a product session ends, there is new information from OpenAI, Anthropic, or a dozen other frontier model builders that could affect what customers expect. Keeping a real-time pulse on that and integrating it into roadmap decisions is becoming a continuous responsibility that no founder or CEO can own alone.
10. The weak correlation between shipping volume and value creation is one of Emrecan’s core concerns about how AI is changing product work. It is now relatively cheap and predictable to ship features. What is hard is demonstrating that what you shipped created net new value rather than just redistributing existing engagement. A feature that gets adopted in one area while another feature drops in usage may represent zero net gain for the customer. He pushes his team to think in terms of system-level impact, not feature-level output.
11. Glean’s product org faces a tension that shapes almost everything about how it is designed. Glean serves the world’s largest enterprises, which expect security, stability, and no surprises. Customers create 60-page internal enablement decks to train their employees on how to use Glean, which means any unexpected change can create real organizational friction. At the same time, those same customers expect Glean to ship at the frontier and stay ahead of competition. Managing those two forces, maximum predictability and maximum pace, is what Emrecan describes as one of the central arts of running product at Glean.
12. His answer to that tension is a differentiated autonomy model. Some teams have almost infinite autonomy to define their roadmap and ship without seeking approval. Other teams operate in a mode where nothing ships without sign-off from Emrecan or the co-founders, because the risk of a surprise to customers is too high. Autonomy at Glean is not assigned. It is taken, by demonstrating the judgment and customer understanding to earn the trust that comes with it.
13. Emrecan has roughly 28 direct reports and almost no recurring one-on-ones. His rationale is that filling a calendar with weekly one-on-ones means that when someone really needs him for a critical decision, there is no room. He keeps the calendar deliberately open so that meetings happen when there is a genuine reason, not because they are on a schedule. He acknowledges this model is not for everyone, and that it creates a high expectation of independence from the moment someone joins.
14. The org is designed with intentional flatness. There is only one layer of PM management between Emrecan and the most junior PM hired. When PMs do manage others, they manage very small teams, typically two to four people, all with the same expectation of autonomy. The question he keeps asking is how long they can sustain that structure before needing to add another management layer, and he is actively trying to extend it as long as possible.
15. Colocation turned out to matter more than expected. When Glean opened a second office in San Francisco alongside its Palo Alto headquarters, EPD pods that had engineers in one office and a PM or designer in the other lost the daily high-touch rhythm that made them effective, including the ability to spot a bug at 1:30 pm and ship a fix at 1:35 pm. The correction was to anchor EPD teams to a single location, because that kind of real-time responsiveness only works when the team is in the same physical space.
16. Glean is, in Emrecan’s description, a meta-product for the company itself. The same enterprise AI assistant they sell to customers is the operating system for how Glean runs internally. Organizational intelligence about what people are working on, which product reviews got attention, and which decisions were escalated or left open is continuously tracked and accessible without anyone having to write a status update. Emrecan can ask Glean what one of his PMs worked on last week without that PM having published anything explicitly.
17. He uses Glean to prepare for one-on-ones in real time. Before a meeting with a PM, Glean surfaces the most likely agenda items based on what that PM has been working on, what customers have been saying about their product area, and what strategic priorities are relevant at that moment. He does not need to manually prepare. That capability, he argues, is what makes it possible to have 28 direct reports and still show up to each conversation with the right context.
18. The tension between consumer-product speed and enterprise-product predictability shapes how he thinks about what product management actually is at Glean versus what it would be somewhere else. He is explicit that the Glean model, including the autonomy expectations, the colocation priority, and the light management structure, is specific to Glean’s situation. A consumer company with different customer expectations would require a very different sliding scale on almost every dimension.
19. On the question of what makes someone a rising PM rather than a plateauing one, Emrecan points to systems thinking as the most indispensable skill. Not just understanding cause and effect within a single feature, but understanding how a change in one part of the product creates downstream and upstream effects across the entire system. The PMs who can see that whole graph, and who optimize for the system rather than for the individual feature, are the ones who create disproportionate impact.
20. His synthesis of where the profession is going comes back to the combination of judgment, systems thinking, and taste. AI is making the executional parts of PM work cheaper and faster. What that does is make the parts AI cannot do more valuable in relative terms. Deciding what deserves to be built, seeing where the real bottlenecks are, understanding what a customer actually needs versus what they say they want, and maintaining quality and coherence across a portfolio of fast-moving work: those are not skills that get automated. They are the skills that define the next generation of product leaders.