Skip to main content

What Actually is an AI Product Manager?

So you've probably seen job postings for "AI Product Manager" or "AI PM" lately, right? Ever since ChatGPT shook up the world, pretty much every company seems to be hunting for AI PMs. But here's the thing - when you actually ask someone "What exactly is an AI PM?", getting a clear answer isn't as easy as you'd think.

I was confused at first too, honestly. I'd been working in this field under the "Intelligence" banner even before LLMs became a thing, but I wasn't really sure when I officially became an "AI PM." I kept asking myself, "Am I actually an AI PM right now?"

So today, I wanted to share my honest thoughts about what an AI PM really is and clear up some common misconceptions, based on my experience working in this field from the Intelligence days through to the AI era. I hope this helps anyone who's dreaming of becoming an AI PM or is just curious about the role.

How I Became an AI PM

My first product was a service centered around "Contextual Suggestion" as its main feature. It extracted various user contexts and curated relevant information to provide to users. Was I an AI PM back then?

To be completely honest, probably not in the strict sense. The context extraction technology running in the background was rule-based. I was defining specific conditions and determining what information to provide under those conditions - basically writing requirements for every condition and action. That was pretty much the classic Product Manager role.

The second product I worked on was a Voice Assistant. It was a product that included understanding user intent using voice technology and deep learning. Technically, you could call this an AI Product. But the features I was responsible for weren't based on AI technology, so personally, I don't think I could call myself an AI PM at that point.

The real turning point came after LLMs emerged. When we started integrating LLMs into the Voice Assistant product, I became responsible for RAG (Retrieval-Augmented Generation) based features. That's when I think I truly started working as an AI PM.

Why? Because it required a completely different kind of planning than before. I had to design "intelligent" behaviors that couldn't be defined by clear rules, understand the characteristics and limitations of AI models to design features, and adjust AI behavior through prompt engineering.

Currently, I'm not working at the app or service level anymore - I'm now responsible for the AI model itself. I've become what you'd call a typical AI PM. I define what capabilities the model needs, discover and package new abilities that emerge from LLM characteristics to make them usable, and continuously think about what direction LLMs should take through what technologies, and how to differentiate from competing models.

Common Misconceptions About AI PMs

One thing I've realized through my experience is that many people have misconceptions about AI PMs. The biggest one seems to be the belief that "only PMs who work on models, not apps or services, are AI PMs."

But I don't think that's accurate. Even if you're working on apps or services, if you're dealing with AI technology, I'd say you can be considered an AI PM. For example, tons of apps these days are adding features that utilize LLMs, right? I think PMs who plan these features can also be considered AI PMs. They're testing AI capabilities, evaluating whether they're viable, and planning features and writing requirements based on those assessments.

Another misconception is that you need to be an AI technology expert to become an AI PM. I don't think that's true either. Sure, you need to understand AI, but you don't need to know every technical detail. What's more important is the ability to identify what AI can and cannot do, and based on that, develop realistic product strategies. Communication with the researchers who actually train the models is more crucial.

Two Types of AI PMs

Looking back at my experience, I think AI PMs can be broadly divided into two categories.

The first is the AI Model PM. When I say AI model, I'm not just talking about LLMs. This includes LLMs, LMMs (Large Multimodal Models), LWMs (Large World Models), and more specifically, speech recognition models, speech generation models, and models based on deep learning or machine learning. AI Model PMs add requirements to these models and provide direction on what capabilities they should offer. Based on this, they enable researchers to train the models.

The second is the AI Experience PM. They add new features or create new experiences in existing products by utilizing AI. AI Experience PMs need to define current problems and determine whether these problems are suitable to be solved using AI.

Core Competencies of an AI PM

After doing this work for several years, I've realized that AI PMs need some special competencies beyond typical PM skills.

First is prompt engineering. Using AI seems to be a fundamental competency. It's not just about asking questions - you need to be able to write prompts that can extract exactly the answer you want from AI. This is harder than it sounds. "Analyze user reviews" and "Classify the following user reviews with sentiment scores from 1-5, extract 3 main keywords for each score, and organize them in JSON format" can produce completely different results.

Second is the ability to evaluate AI. It's really important to objectively assess AI and establish criteria for its use. AI sometimes shows amazing results, but it can also give you completely wrong answers. You can't just stop at "Wow, this is incredible" - you need to constantly question how reliably it will work in an actual service, in what situations it might fail, and whether the quality is acceptable to users.

Third is strategic planning ability. You need to define direction and enable the development of AI models based on that, and establish strategies for differentiation in an era where countless AIs are emerging. It's not just about "Let's add features like ChatGPT too," but continuously thinking about what AI experiences our users really need and what differentiation is possible with our data and technology.

Finally, there's communication ability with researchers and developers. You need to be able to communicate what problems to solve with AI and what features to provide, and play a role in ensuring optimal models are developed based on that. In some ways, this is in the same context as the competencies needed for traditional PMs.

Insights

Through these experiences, I've gained several insights.

First, AI PM isn't a completely new role but an evolved form of the existing PM role. The core is still solving user problems, and AI is just a new tool for solving those problems. It's just that this tool is so powerful and special that it requires separate expertise.

Second, a sense of balance is really important. Both excessive expectations and too much skepticism about AI can be dangerous. You need to believe in AI's potential while acknowledging its realistic limitations, keep pace with the speed of technological advancement while focusing on users' actual needs.

Finally, an AI PM needs to be someone who continuously learns. This field is changing so rapidly that what was common knowledge a few months ago often doesn't apply anymore. When a new model comes out, you need the curiosity to try it yourself and the ability to think about how to apply new techniques when they emerge.

For Those Dreaming of Becoming an AI PM

If you're interested in becoming an AI PM, I don't think you need to start with something grand. How about looking for small opportunities to utilize AI in your current work? Just using ChatGPT to improve work efficiency or doing simple prompt experiments can be a great starting point.

And most importantly, I hope you don't think you need to give up everything and start fresh to become an AI PM. The PM experience and domain knowledge you've built up will remain powerful assets. If you just add understanding and utilization of AI to that foundation, I think you can become a really strong AI PM.

The AI era seems to have just begun. I'm curious too about how AI PMs will continue to evolve from here.

Popular posts from this blog

5 Game-Changing Ways X's Grok AI Transforms Social Media Algorithms in 2026

5 Game-Changing Ways X's Grok AI Transforms Social Media Algorithms in 2026 In January 2026, X (formerly Twitter) fundamentally reshaped social media by integrating Grok AI—developed by Elon Musk's xAI—into its core algorithm. This marks the first large-scale deployment of Large Language Model (LLM) governance on a major social platform, replacing traditional rule-based algorithms with AI that understands context, tone, and conversational depth. What is Grok AI? Grok AI is xAI's advanced large language model designed to analyze nuanced content, prioritize positive and constructive conversations, and revolutionize how posts are ranked and distributed on X. Unlike conventional algorithms, Grok reads the tone of every post and rewards genuine dialogue over shallow engagement. The results are striking: author-replied comments now receive +75 ranking points —150 times more valuable than a single like (+0.5 points). Meanwhile, xAI open-sourced the Grok-powered algorithm in Ru...

How Claude Opus 4.6 Agent Teams Are Revolutionizing AI Collaboration

Imagine delegating complex tasks not to a single AI, but to a coordinated team of specialized AI agents working in parallel. Anthropic's Claude Opus 4.6, unveiled on February 5, 2026, makes this reality with Agent Teams —a groundbreaking feature where multiple AI instances collaborate like human teams, dividing roles, communicating directly, and executing tasks simultaneously. As someone deeply engaged with AI systems, I found this announcement particularly compelling. Agent Teams represent a fundamental shift from solitary AI execution to collaborative multi-agent orchestration, opening new possibilities for tackling complex, multi-faceted problems. How AI Agent Teams Actually Work The architecture of Agent Teams is surprisingly intuitive—think of it like a project team in a company. At the top sits the Team Lead , an Opus 4.6 instance that oversees the entire project, breaks down tasks, and coordinates distribution. Below the Lead are Teammates , each running as indepen...

AI Agents Hit Reality Check: 5 Critical Insights from the 2026 Trough of Disillusionment

AI agents are everywhere in 2026. Gartner predicts 40% of enterprise applications will embed AI agents by year-end—an 8x jump from less than 5% in 2025. But here's the uncomfortable truth: generative AI has already plunged into the "Trough of Disillusionment," and AI agents are following the same path. While two-thirds of organizations experiment with AI agents, fewer than one in four successfully scales them to production. This isn't just another hype cycle story. It's a critical turning point where ROI matters more than benchmarks, and the ability to operationalize AI determines winners from losers. The Hype Cycle Reality: Where AI Agents Stand in 2026 According to Gartner's Hype Cycle for AI 2025, AI agents currently sit at the "Peak of Inflated Expectations"—the highest point before the inevitable crash. Meanwhile, generative AI has already entered the Trough of Disillusionment as of early 2026. What does this mean for enterprises? Gartner fo...