Sign up to save your podcasts
Or
Share #017: Product Management in 2025: What Broke, What Stayed, What Matters
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
#017: Product Management in 2025: What Broke, What Stayed, What Matters
Seven months ago, we talked about how product management was changing. We focused on discovery over delivery and outcomes over outputs. These topics felt urgent then. Now they’ve faded into the background. AI has taken center stage. And this shift is justified.
Canva disrupted design. Shopify disrupted e-commerce. AI is doing the same to product management. The way product management was done six months ago will not exist in the future. At least not in most companies.
The PM as builder
AI coding agents have become so good that every product manager needs to evolve. The shift goes from facilitator and communicator to builder. Not just building prototypes to validate. Actually touching code and shipping features.
This was only enabled in the last two or three months. Before that, AI wasn’t capable of it. Now even semi-technical people can ship features on their own.
Six months ago, most PMs used AI for writing and discovery. Bouncing ideas back and forth with ChatGPT. Now we spend hours in tools like Lovable or Claude Code. Creating things that go beyond clickable prototypes.
The old way: Sit with designers in Figma. Create linear prototypes where only one button works. A yellow highlight shows which element is clickable. This took three weeks.
The new way: Have an idea in the morning. By afternoon, have something that feels like production. Everything is clickable. Actions happen in the background. Hand it to a developer and they can turn it into production code in a fraction of the time.
The death of the feature delivery pipeline
Claude Code blew me away when I started using it. It feels optimized for product managers. Like working with a senior engineer who checks in at the right time.
When planning a new feature, it spends 10-15 minutes exploring the code base. Finding out what needs to be touched. Then it launches a planning agent. Comes up with a comprehensive plan for review.
This is exactly how great engineers worked before. They surfaced risks. They surfaced increased costs. They suggested what should be a follow-up feature versus what to build now.
The old feature delivery pipeline looked like this: PM talks to customers. Forms an idea. Prioritizes it. Does a first check-in with UX. Works on wireframes for two weeks. Checks feasibility with the tech lead. It doesn’t work. Back to the drawing board. Six weeks of coordination overhead. Then six weeks of implementation.
This is already outdated.
Rethinking the feature delivery pipeline will be one of the biggest challenges for organizations in 2026.
Teams have to change
In a company with eight engineers per product manager, the PM can only be a facilitator. A communicator. Someone who makes sure everyone works toward the same thing. If engineers get faster with AI tools, this coordination work only increases.
Andrew Ng, founder of Coursera, recently suggested his team is moving to two product managers per engineer. More PMs than engineers. Six months ago, this would have sounded insane.
But delivery is no longer the bottleneck.
Before, there was a clear cut. PMs focused on discovery. Engineers focused on delivery. Sometimes in forward-thinking organizations, engineers did some discovery too. But PMs rarely touched delivery.
Now the lines blur. While doing discovery, you can already deliver. Talk to a customer in the morning. Understand an insight. Prototype something during the call. Polish it in the afternoon. Push to production by evening.
Discovery and delivery happen in the same motion.
The risk of feature soup
If everyone can deliver quickly, the risk of Frankenstein platforms increases. Products that don’t make sense anymore. Features piled on features without a clear thread.
Organizations without a clear vision will have a hard time if they enable everyone to be builders. Products need to become more specialized. The old economics didn’t justify building a product with an ICP of 10,000 people. Now it does.
But this requires discipline. Stick to a strategy. Don’t pivot every quarter. Otherwise you pile up features that become hard to maintain. Not just technically. Also organizationally, contractually, and when onboarding new people.
Feature bloat without strategy is one of the biggest risks ahead.
Distribution becomes the differentiator
Before AI, success required a pyramid. The lowest level: ability to build. You needed engineers or money. This level is completely gone.
Second level: having a great product. Simple, solving the right problems, intuitive, accessible.
Third level: distribution. This was always there. But it’s becoming the main differentiator.
What Shopify did to e-commerce is what AI will do to software. Building a good product is no longer a competitive advantage. It’s the ultimate baseline.
A fitness instructor working at 10 different fitness centers has access to 200 potential customers. That’s an unfair advantage. They can test their app. If it’s good, word of mouth spreads.
The skills that will matter for product managers: networking, sales, messaging, and marketing. When aligning with engineers about feasibility becomes less relevant, focus shifts to business fundamentals.
The ability to create something great is not as valuable anymore. It was very valuable a year ago.
What’s next
Predicting the future is hard. Seven months ago, the rate of change felt crazy. It was even crazier than expected. Small improvements compound into fundamental shifts.
Prompting was the hot topic six months ago. Now the hottest topic is keeping the most relevant context without having to retype it. In six months, we’ll probably laugh about manually attaching screenshots.
People overestimate what can be achieved in the short term. They underestimate what can be achieved in the long term.
A lot has changed. But it has never been more exciting to work in product and software.
Links
Link to Podcast Episode
* 📹 YouTube
* 🔊 Spotify
* 🔊 Apple Music
In case you want to reach out, please do so on LinkedIn:
* ❤️🩹 Follow Hotfix: https://pal.bio/the-hotfix-podcast
* 🎙️ Follow Christoph: https://www.linkedin.com/in/christophbodenstein/
* 🎙️ Follow Stefan: https://www.linkedin.com/in/stefan-pernek-629901107/
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit thehotfixpodcast.substack.com
View all episodes
By The Hotfix PodcastSeven months ago, we talked about how product management was changing. We focused on discovery over delivery and outcomes over outputs. These topics felt urgent then. Now they’ve faded into the background. AI has taken center stage. And this shift is justified.
Canva disrupted design. Shopify disrupted e-commerce. AI is doing the same to product management. The way product management was done six months ago will not exist in the future. At least not in most companies.
The PM as builder
AI coding agents have become so good that every product manager needs to evolve. The shift goes from facilitator and communicator to builder. Not just building prototypes to validate. Actually touching code and shipping features.
This was only enabled in the last two or three months. Before that, AI wasn’t capable of it. Now even semi-technical people can ship features on their own.
Six months ago, most PMs used AI for writing and discovery. Bouncing ideas back and forth with ChatGPT. Now we spend hours in tools like Lovable or Claude Code. Creating things that go beyond clickable prototypes.
The old way: Sit with designers in Figma. Create linear prototypes where only one button works. A yellow highlight shows which element is clickable. This took three weeks.
The new way: Have an idea in the morning. By afternoon, have something that feels like production. Everything is clickable. Actions happen in the background. Hand it to a developer and they can turn it into production code in a fraction of the time.
The death of the feature delivery pipeline
Claude Code blew me away when I started using it. It feels optimized for product managers. Like working with a senior engineer who checks in at the right time.
When planning a new feature, it spends 10-15 minutes exploring the code base. Finding out what needs to be touched. Then it launches a planning agent. Comes up with a comprehensive plan for review.
This is exactly how great engineers worked before. They surfaced risks. They surfaced increased costs. They suggested what should be a follow-up feature versus what to build now.
The old feature delivery pipeline looked like this: PM talks to customers. Forms an idea. Prioritizes it. Does a first check-in with UX. Works on wireframes for two weeks. Checks feasibility with the tech lead. It doesn’t work. Back to the drawing board. Six weeks of coordination overhead. Then six weeks of implementation.
This is already outdated.
Rethinking the feature delivery pipeline will be one of the biggest challenges for organizations in 2026.
Teams have to change
In a company with eight engineers per product manager, the PM can only be a facilitator. A communicator. Someone who makes sure everyone works toward the same thing. If engineers get faster with AI tools, this coordination work only increases.
Andrew Ng, founder of Coursera, recently suggested his team is moving to two product managers per engineer. More PMs than engineers. Six months ago, this would have sounded insane.
But delivery is no longer the bottleneck.
Before, there was a clear cut. PMs focused on discovery. Engineers focused on delivery. Sometimes in forward-thinking organizations, engineers did some discovery too. But PMs rarely touched delivery.
Now the lines blur. While doing discovery, you can already deliver. Talk to a customer in the morning. Understand an insight. Prototype something during the call. Polish it in the afternoon. Push to production by evening.
Discovery and delivery happen in the same motion.
The risk of feature soup
If everyone can deliver quickly, the risk of Frankenstein platforms increases. Products that don’t make sense anymore. Features piled on features without a clear thread.
Organizations without a clear vision will have a hard time if they enable everyone to be builders. Products need to become more specialized. The old economics didn’t justify building a product with an ICP of 10,000 people. Now it does.
But this requires discipline. Stick to a strategy. Don’t pivot every quarter. Otherwise you pile up features that become hard to maintain. Not just technically. Also organizationally, contractually, and when onboarding new people.
Feature bloat without strategy is one of the biggest risks ahead.
Distribution becomes the differentiator
Before AI, success required a pyramid. The lowest level: ability to build. You needed engineers or money. This level is completely gone.
Second level: having a great product. Simple, solving the right problems, intuitive, accessible.
Third level: distribution. This was always there. But it’s becoming the main differentiator.
What Shopify did to e-commerce is what AI will do to software. Building a good product is no longer a competitive advantage. It’s the ultimate baseline.
A fitness instructor working at 10 different fitness centers has access to 200 potential customers. That’s an unfair advantage. They can test their app. If it’s good, word of mouth spreads.
The skills that will matter for product managers: networking, sales, messaging, and marketing. When aligning with engineers about feasibility becomes less relevant, focus shifts to business fundamentals.
The ability to create something great is not as valuable anymore. It was very valuable a year ago.
What’s next
Predicting the future is hard. Seven months ago, the rate of change felt crazy. It was even crazier than expected. Small improvements compound into fundamental shifts.
Prompting was the hot topic six months ago. Now the hottest topic is keeping the most relevant context without having to retype it. In six months, we’ll probably laugh about manually attaching screenshots.
People overestimate what can be achieved in the short term. They underestimate what can be achieved in the long term.
A lot has changed. But it has never been more exciting to work in product and software.
Links
Link to Podcast Episode
* 📹 YouTube
* 🔊 Spotify
* 🔊 Apple Music
In case you want to reach out, please do so on LinkedIn:
* ❤️🩹 Follow Hotfix: https://pal.bio/the-hotfix-podcast
* 🎙️ Follow Christoph: https://www.linkedin.com/in/christophbodenstein/
* 🎙️ Follow Stefan: https://www.linkedin.com/in/stefan-pernek-629901107/
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit thehotfixpodcast.substack.com