How has Product Management Changed with AI?
The History of AI in Product Management (2023-2025)
I'll quickly cover what's happened in the last couple of years with AI in Product Management. There's quite a lot of companies hyping Product Management as being hugely transformed by AI, but my discovery conversations reveal that's not actually the case (outside of tools like claude).
2023
In 2023, generative AI tools burst onto the scene, creating what McKinsey called a "breakout year" for AI adoption. Their research found that one-third of organizations began using generative AI regularly in at least one business function, with product development being among the top three application areas. However, only 21% of companies had established governance policies for AI use, highlighting the experimental nature of this phase.
During this period, product teams primarily used AI for:
Automating routine documentation tasks
Generating initial drafts of product requirements
Basic customer feedback analysis
2024
In 2024, AI tools became more specialized for product management workflows. Forbes wrote, "AI's integration into product management is just the beginning".
Key developments in 2024 included:
Purpose-built AI product management platforms like Rough
AI-powered (but functionally useless) "Chief Product Officer" copilots that could transform brief outlines into full PRDs
The emergence of specialized AI education for product managers, including Udacity's "AI Product Manager Nanodegree"
2025
In 2025, there's been a fundamental shift toward "AI-first strategies" in product management. This represents a move beyond treating AI as a feature to making it the foundation of product strategy and operations. Most of this pressure has come from VCs and CEOs, wanting to stay ahead of the curve.
The current state of AI in product management is characterized by:
Predictive roadmapping that uses AI to forecast market needs
Real-time customer insight analysis that informs product decisions
A broader understanding across the industry of what AI is and isn't good at
"Product management is now based on real-time, predictive insights thanks to artificial intelligence rather than intuition and historical data," - The only research we could find on this topic.
This stuff sounds like most "traditional" PMs are being left behind, but I would caution against those sorts of feelings. I think most people aren't publishing articles on "Not much has significantly changed since 2023" so the majority of sources we've found for 2025 have overstated the impact of AI.
AI Product management Articles
This page is a general overview of the state of Product Management in 2025. If you're looking for more specific information, check out some of these articles.
The Current State of AI in Product Management (2025)
If you're a PM trying to figure out what to do with AI in 2025, you're not alone. The tech has moved fast, and it's easy to feel overwhelmed.
First things first: AI is becoming a larger part of how most companies work. Most top-performing companies are using it in some way or another.
What's working for PMs right now
The most successful product teams are using AI in three main ways:
They're letting AI handle the boring stuff. AI tools can now write your release notes, summarize feedback, and even draft basic requirements. This gives you back hours each week to focus on the work that actually needs your human touch.
They're focusing on fewer, better features. Instead of building lots of okay features, AI helps teams identify which few capabilities will make the biggest difference. It's quality over quantity.
How to get started
You don't need to transform everything overnight. Here's a simple way to begin:
1. Pick one task you hate doing. Maybe it's writing status updates or organizing feedback. Find an AI tool that can help with just that one thing.
2. Ask yourself: "What decisions am I making with too little information?" AI is great at analyzing more data than humans can handle. If you're guessing about user needs or market trends, that's a perfect place for AI to help.
3. Start small with a team experiment. Try an AI tool for two weeks and see if it actually helps. Not every shiny AI feature will work for your specific needs.
What you don't need to worry about
Despite what you might hear, you don't need to become a technical AI expert. You need to understand what AI can do, not how to build it. You also aren't going to replace your entire workflow at once. The best teams are adding AI tools gradually where they make the most sense.
Finally stop worrying that AI will replace you. The most valuable PM skills – empathy, strategic thinking, and cross-team collaboration – are exactly what AI can't do well.
What's actually working
You can be making roadmap decisions with more confidence. AI helps process more market data, competitive intelligence, and user feedback than any human team could – leading to better bets on what to build next.
AI works best when it handles the things machines do better than humans (processing data, spotting patterns, automating routine tasks) while freeing you up to do what humans do best – understanding the "why" behind user needs, making creative connections, and building relationships across teams.
So start small, focus on real problems, and remember that AI is just a tool to help you build better products – not a replacement for the human judgment that makes you valuable as a PM.
Making Product Decisions for AI Features
Ok what about if you're tasked with adding AI to your own product? First things first, start with the problem, not the technology. Before you add AI to your product, ask yourself: What user problem will this solve? The best AI features address specific pain points that couldn't be solved well without AI. For example, instead of saying "Let's add AI recommendations," start with the problem of "Users are having a tough time discovering content"
Set realistic expectations
Be honest about what AI can and can't do right now. Many teams get in trouble by promising perfect AI that never makes mistakes. Instead, think about how to design features that are still valuable even when the AI isn't perfect. This might mean showing confidence scores, giving users ways to correct mistakes, or designing the interface to make AI limitations clear.
Consider the full user experience
AI features don't exist in isolation. Think about how they fit into your users' workflow. A great AI feature that takes users out of their flow might actually hurt the overall experience. Think about when and how the AI should appear, how much control users need, and what happens when they want to override or adjust what the AI is doing.
Plan for feedback loops
The best AI features get better over time. Design ways to collect feedback on AI performance from the start. This could be explicit feedback (thumbs up/down buttons) or implicit signals (did the user accept or ignore the AI's suggestion?). Make sure your team has a process for reviewing this feedback and improving the model.
Build trust gradually
Users need time to build trust with AI features. Start with low-risk applications where mistakes won't frustrate users, then gradually expand as you earn their confidence. For example, you might start with AI that suggests actions rather than taking them automatically, then add options for automation as users get comfortable.
Address ethical concerns early
Every AI feature comes with ethical questions. Who benefits from this feature? Could it harm or exclude certain users? How transparent should we be about how it works? These aren't just nice-to-have considerations – they're essential product decisions that affect adoption and trust. Make ethics part of your product discussions from day one.
Remember that you're not trying to add AI everywhere – you're just trying to user problems in the best way possible. Sometimes that means sophisticated AI, sometimes it means a simple non-AI solution, and often it means a thoughtful combination of both. The products that win aren't the ones with the most advanced AI; they're the ones that use AI to deliver experiences that truly help their users succeed.
Product Management Articles
The rest of our articles on Product Management, with a focus on AI.
Rough
Join our slack for product updates, and discussions with the Rough team.
Alternatively, you can reach out to us directly at hello@rough.app