There have been dozens of hot takes in response to Replit CEO, Amjad Masad, saying that he doesn’t think people should learn to code. While this may be considered another such take, I’m actually much more interested in the alignment AI causes with product- and engineering-capable people. So, I guess, let’s get this out of the way up front. Should you still learn to code?

Yes.

With that out of the way, let’s explore something more interesting. Product management as it has existed—and it has only had a very short life—is dead. But to be fair, AI didn’t kill it. In fact, AI is likely its savior (more on that in a minute). Product management has been on its death bed since organizations began to realize the value of technical product management which gave way to engineers doing product research and development. In many cases, product management never got the chance to exist properly because PMs were treated as if they were project managers.

Even in organizations where product management was taken seriously, it has become evident that coordinating with developers when you are non-technical results in a reduction back to project management. The highest performing product and engineering teams seem to have a common trait in that they have product managers with an engineering background, or they have engineers doing what would amount to product management.

Maybe this is a good time to step back and define product management.

What is product management?

Product management is hard to define, but I like to think of it in analogies. So pick your analogy and skip the ones you don’t like.

Product managers are like conductors in an orchestra—they don’t play the instruments, but they know enough about the instruments to help those who do play perform at their best.

Product managers are like a racing pit crew—they aren’t driving the car, but they know as much about the car as the driver and can guide the driver through the race.

Product managers are like film directors—they don’t act in the scenes or operate the cameras, but they have the vision for the final product, coordinate diverse specialists, make critical decisions about what stays or goes, and ultimately take responsibility for whether the project succeeds or fails.

Once you’ve picked your metaphor and have a perfect understanding of what product management is, let’s define what it is not. Product management is not a role that carries any real power. Influence, sure. Power, no. A product manager can’t fire people, generally. A product manager can’t MAKE something happen. They are politicians, but with less power. They are diplomats, but with more knowledge.

That might be the last meme of this article, but no promises.

Product management is dead

Back to my original point. Product management was designed to help engineers build the right thing for the right people at the right time. However, engineers speak a different language than many product managers. So, many product development teams realized that they could put engineers and former engineers into this role and see much better results. When you don’t have to translate between normie speak and engineer speak, the process is much faster. Trust me, I learned this over time as I became more technical.

I started out in the tech world working at a massive insurance company on their claims app. I didn’t know how to code, but I had a basic understanding of tech. My job was to translate the results of my research into requirements documents engineers could understand. This was a steep learning curve. In my next job, I worked for a startup and it was there that I learned the value of understanding code and engineering practices as part of the product roadmap, prioritization, and release flow. Eventually, I did learn to code, worked as a developer at a few places, and then leveraged that experience to propel my product career forward.

So, it came as a surprise when I took a short detour and worked for a larger, pre-IPO startup who had a whole team of product managers who were non-technical. I saw the differences in-person and as stark as day and night. Communication is filled with much more friction when product managers and engineers don’t have a shared language and shared understanding of technical capabilities.

If you follow early-stage startups, you’ll see a lot of these companies focused on engineering hires only. The stated reason is efficiency, but I think it’s actually a solution to communication problems, which of course translates to better efficiency.

So, product management is dead, right?

Long live product management

Product management is not dead, and you should learn to code, and engineering as we know it will be changed forever. All of these things appear to be true at once.

In a recent episode of Lenny’s Podcast, Lenny interviewed Eric Simmons, the CEO of Stackblitz. You might not know Stackblitz, but you probably know Bolt, their AI coding app that allows you to build entire projects in the browser using AI. In this interview, both Lenny and Eric make predictions about the future of engineering and product management. One prediction stood out.

They both agreed that product management will become more important than ever in the age of AI. In fact, they suggest without explicitly stating this that product management will become more important than engineering. But it will require a certain type of product manager.

Creating apps with AI requires a type of prompting that should feel familiar to product managers. In normal product management, you might write a high-level document or you might get detailed with a full PRD (product requirements document). Prompting AI to code uses the same methods. Go broad and let the AI be creative, or get detailed and make the AI be specific. Then, just like in product management, there’s a back-and-forth review and adjustment period.

This image is nonsensical and hilarious. I asked ChatGPT’s new image generation model to represent what I said above and this is what it came up with. Unlike a good PM, I didn’t ask for any revisions. ¹⁰⁄₁₀ no notes. But I digress…

The skills necessary to prompt AI to build the right thing for the right people is a direct descendant of product management. But there’s an extra step that will make product managers superhuman as AI continues to take hold. That step is learning to code.

But wait, if AI is doing the coding, why should the person prompting have to learn how to code? This is pretty simple actually, and it’s the exact same reason many startups are turning to technical, former-engineer, product managers today. If you understand the code, if you understand what’s possible and what’s not, you can correct and guide an LLM significantly better.

This creates a chicken and the egg problem if you’re not careful. If you need to learn to program to be a good AI-age product manager, how do you learn to program in a world where AI is probably doing the teaching. The answer, to me at least, is simple. You prompt some more. Prompting is conversation. Prompting is asking questions and following up. Prompting is testing theories. Prompting is what humans have done forever, we just called it talking.

And traditional ways of learning to program aren’t going to just dry up. There will be courses on YouTube, blog posts, and—can you believe it?—books. But, in my very biased opinion, the best way to learn how to code is to build things for yourself. Do that as many times as it takes for you to start understanding the conversation your engineering friends are having. Do it as many times as it takes to start catching logical issues in AI outputs when you prompt it to build something.

This future, powered by AI, is much more likely to be populated with more product managers, not fewer. And the ones that make it will have the coding skills necessary to execute in an environment where AI does a good chunk of the coding itself.