Unlocking the AI PM Dream: Your Roadmap to Success
Unlocking the AI PM Dream: Your Roadmap to Success
What It Really Takes to Be an AI PM
I have something to admit: I’ve never taken a class on statistics (hardly picked up a book on this). I’ve never even taken a course on Machine Learning.
Whether you're building AI products or just fascinated by the field, you've probably noticed the surge in demand for AI Product Managers. As someone who's been navigating the tech industry specifically in AI for years (from working on evaluation for self driving cars, to powering recommender systems at Spotify), I've had a front-row seat to this AI revolution.
How did I end up working with some of the leading AI engineers building products we all know and love?
I’m going to cover how you can break into this role, starting from zero:
What companies to look at - these will be shared out regularly so feel free to subscribe for the latest here
Step by step, what skills do you need to succeed and thrive as an AI PM, with an example each week
Lessons from other AI PMs
I know how to do this because I pivoted into an AI PM role myself.
I started my career as a Mechanical Engineer, and retooled myself for AI software in the early stages of the AI/ML wave (around 7 years ago now), before pivoting into an AI PM role across Data Science and Engineering. I realized there was a ton of lessons along the way I wish others had shared with me - like how to get started, or whats real vs noise. I want to share more of the frameworks, resources and lessons I’ve learned, in the hopes that it helps you land that dream AI job - and eventually thrive as an AI PM.
We need more builders!
But before we jump into how to get that AI dream job - it’s important to realize that AI is a wave just like any other prior wave in tech.
I’d say this more recent wave of AI has a significant amount more exuberance, but its likely that the role of an “AI PM” actually becomes just like any other “PM”. AI should become more and more a part of ANY product we build or consume in our lives, so the principles you learn for becoming an AI PM should be transferable to building products in any domain. If you’re already a PM, knowing the ins and outs of AI systems is going to help you succeed in building products that push the boundaries of what is possible with technology.
Now, let's break down what AI PMs actually do
AI Product Managers are the connective tissue between technical AI teams (engineering, builders) and business objectives (GTM, sales, customer facing teams). They come in a few different flavors:
AI Powered PMs: These PMs integrate AI into existing products. Imagine working on Gmail's Smart Compose feature or Spotify's recommendation engine. In our case, this is our AI Copilot. The challenge here is seamlessly blending AI capabilities into user experiences in a way that feels natural and adds real value.
AI Product PMs: These are PMs that manage products where the Model is the product (i.e. chatGPT). The role is in between AI Platform PMs for research organizations (i.e. DeepMind, OpenAI, Meta AI research) and shipping AI powered experiences on top of the model. They might be exploring how to use AI for drug discovery at a biotech company, or developing next-gen language models at OpenAI. The goal is to push the boundaries of what's possible and turning theoretical breakthroughs into practical applications.
AI Platform PMs: These folks focus on building tools for other developers. This is a lot of where my thinking goes at Arize - we’re building Open Source tools to help builders understand what their AI products are actually doing, and how they work. Think of products like Langchain/Llamaindex. Their job is to make AI development more accessible and efficient for other engineers. It's all about creating the right abstractions and interfaces that make complex AI operations feel intuitive.
Now, you might be thinking, "Okay, but how do I actually land one of these roles?" Well, that's where things get interesting. It's not just about having the right skills - it's about demonstrating them in ways that make you stand out. We'll dive into that in a bit.
The key thing to remember is that regardless of the specific flavor, all AI PMs share a common challenge: bridging the gap between complex technical possibilities and real-world user needs. This is a balancing act that requires both technical chops and a keen understanding of product strategy.
Let's dive in and unpack what it really takes to become an AI PM in today's rapidly evolving tech landscape.
Step 1: Get Familiar with the Tools
You don't need to be the next Geoffrey Hinton to be an AI PM, but you should be able to hold your own in a conversation about neural networks. Here's my playbook for getting up to speed:
Curate your AI product diet - there is a ton of content out there, but picking and choosing the right sources can help you ramp up quickly. Tl;dr AI is a fantastic resource for daily news on AI. For more of the fundamentals, I highly recommend catching up some of Andrej Karpathy’s lectures. Remember not to over-rotate on technical, and learn from amazing product thinkers as well (this is a product role after all!). Lenny’s Podcast and Newsletter is still the canonical source on all things Product - highly recommend checking this out (and a lot of guests give practical advice on approaching AI as well).
Don't just read about LLMs - get your hands dirty. I spent weekends building small projects, breaking things, and learning why they broke. Greg Isenberg has some incredible content here, like getting started with Replit agent to build an AI app.
Obsessively use AI-powered products like you're trying to break them. I became that annoying friend who's always asking OpenAI Voice Mode weird questions or trying to confuse an LLM. It's not just fun - it gives you insight into user experience that you can't get any other way.
Step 2: Demonstrate Your Knowledge
Knowing stuff is great, but in this field, you've got to show it. Here's how I keep up here and what you can do to demonstrate your interests here:
I contribute to a podcast (Deep Papers), where we talk about the latest AI papers in plain English. It forced me to stay current and really understand the tech I was working on.
Side projects are your secret weapon. I built a simple AI that could generate storybooks, with images. Was it revolutionary? No. But it showed I could take an idea from concept to execution, and what I learned in the process made for great conversation on the problems I had to solve along the way.
Share your learning and contribute back - if you’re creating some noise, contributing to open source or just sharing what you are learning, you’re effectively creating a portfolio someone else can follow along with.
Step 3: Connect with the Right People
I’m going to give you 2 hacks here to shortcut straight to getting recognized:
Master the cold outreach by being specific- How many applications come from people who apply actually try the product and show up to the first interview with ideas? Or even better, what if the cold email came with ideas? This gets a lot more attention than an ask for time - it means you’ve spent a bit of time actually getting to understand our problem and challenges, so the discussion will at least be interesting both ways!
Build your network - Linkedin is incredibly powerful. Most importantly, persistently grow your network in the direction of your dream company. Don't just lurk on social media - engage. Spend a lot of time hanging out where people are building. Sometimes, all it takes is asking a thoughtful question on a twitter thread!
Side note: Conferences and meetups are gold mines. I met my current boss at a virtual ML Ops panel, almost 4 years ago! You never know where these connections will lead. You can find your tribe from offline or online communities. It's like having a support group that speaks your language.
Next, we'll dive deeper on the above and pitfalls in the next post, because knowing what not to do is just as important as knowing what to do.