AI Will Take Coders’ Jobs
It’s Not About Juniors, But Those Who Cling to the Past
The twenties of this century are turbulent for the software industry. Pandemics, massive layoffs, the end of the 0% interest rate period, and now AI that is supposed to replace less-skilled engineering workforces.
While the debate over whether the rise of generative LLMs will eliminate our jobs is still ongoing, one thing is certain: a leap forward is ahead of us, and it will be disruptive for many in our field. This is exactly when our roles as engineering leaders become more critical—both for our companies and for our people.
We’re told to fear AI’s imminent takeover—the idea that lines of code will magically write themselves, making engineers obsolete. But the bigger threat lies in how we define success in software teams: Are we churning out code just to “deliver,” or are we building the right products in the first place?
AI Isn’t Replacing Engineers—It’s Changing the Game
Generative AI isn’t just writing code; it’s shining a spotlight on a major flaw: equating more code with more success. For years, organizations ramped up developer headcounts—tens of Project Managers specifying requirements for hundreds of coders who produced millions of lines. But for what?
According to Marty Cagan’s book Transformed, only 10-30% of shipped features actually yield positive outcomes. In other words, piling on more people and writing more code doesn’t necessarily translate to real value. If two-thirds of what we build goes underused, we’re creating engineering waste—whether it’s overly complex solutions or an abundance of unfinished features.
The Data on AI-Driven Coding
So, is AI going to turn code creation into a commodity? Possibly. GitHub reports that tasks can be completed nearly twice as fast with Copilot, while DORA research finds AI adoption correlates with increased flow and productivity—though it also introduces new challenges related to delivery throughput and stability.
No matter where we are now, AI is here to stay. The tools will become better over time, and soon our management practices will also adjust to their existence.
Marty Cagan, in his latest blog post, estimates that current AI-based tools boost engineer productivity by 20-30%. If your team is made up of 6-8 people (1 PM, 1 UI Designer, 4-6 software engineers), that efficiency jump could theoretically compress headcount to 4-6.
Wait, what? Isn’t productivity improvement about delivering more and faster?
Here’s the key: Speed only helps if we’re actually building the right solutions. If your developers spend that saved time just churning out more code for specs handed down by Product Managers, you might see minimal real impact. In some cases, it can bring some disruptions. Just think of all of these coders complaining that “PMs don’t provide enough stories, requirements aren’t clear enough, priorities are changing…”
Productivity was never about how much you can produce but how much real value you can generate. Moreover - from the leader's perspective, it’s not a secret that managing smaller teams is easier. Fewer people mean simplified communication, fewer dependencies, and less managerial work. There is a certain level of delivery (throughput) capabilities above which “more” doesn’t mean “better”.
The Shift from Delivery to Discovery
We’re on the brink of an AI-driven redefinition of software engineering roles. Instead of pumping out 10,000 lines of code that nobody uses, imagine writing 2,000 lines that perfectly solve the right problem. That’s the real future of engineering.
What’s Changing?
Less Coding Required – Tools like Cursor or Copilot can cut coding time by 50% or more.
Smaller, More Focused Teams – Some companies are reallocating resources toward product research, user testing, and validation instead of large squads of coders.
Higher-Level Thinking – AI can draft code, but it can’t define the problem or gauge a feature’s impact. That strategic layer remains distinctly human.
In his last article, Tim O’Reilly says:
AI will not replace programmers, but it will transform their jobs. Eventually much of what programmers do today may be as obsolete (for everyone but embedded system programmers) as the old skill of debugging with an oscilloscope. (…) it is not junior and mid-level programmers who will be replaced but those who cling to the past rather than embracing the new programming tools and paradigms.
If you can finish your programming task 50% faster thanks to AI, asking a PM for another task, pushing them to specify more requirements, or expecting to test your work as a regular user would only create more noise for the team.
The real threat is to developers who refuse to adapt—those who keep equating “lines of code” with “output.”
What This Means for You and Your Team
Pure coding becomes a commodity. It’s analogical situation to Software Testers whose jobs are predicted to decline by at least 15% globally, according to the World Economic Forum’s Future of Jobs Report 2023.
What happened? High-performing organizations—the ones that release continuously or weekly—realised that shifting from human manual work to tests automation allowed to streamline the process of software delivery (I wrote more about it here: Do You Need More Testers or Better Tests?.).
There is another wave of optimization coming to the software delivery process. Instead of just brute-forcing lines of code, today what’s more important is mastering AI-driven workflows, architecture oversight, and product discovery. Engineers writing repetitive boilerplate by hand are missing an opportunity. The real creative work is in validating product ideas and designing better architectures.
The latter is quite clear to everyone. AI code isn’t always flawless. Bugs, security vulnerabilities, and misguided assumptions still happen. Humans are needed to filter and refine AI outputs. Hence, you need strong software engineers—seniors, principals, architects—to ensure that generated code meets standards of scalability, modularity, and performance.
Yet, building the right thing in the first place is even more important than good architecture. A well-crafted distributed system that can be scaled up almost indefinitely in a cloud-native environment simply doesn’t matter if the product doesn't solve customers’ problems.
When code generation becomes increasingly automated, the comparative advantage of human engineers lies in their ability to understand context, define problems clearly, and determine what solutions will truly serve users and the business.
My Own Experience with AI-Driven Discovery
I build CodeAudits.ai, an LLM-powered app for auditing the codebase. Recently, I needed to add functionality in which the user can add their custom API key for the LLM provider of their choice.
I’m not a UI/UX designer, so without much thought, I assumed I’d build a simple RadioGroup with the provider name and an input field for the API key.
In the past, I would spend a week building it only to become frustrated that it wasn’t really what I needed. Today, I can take matters into my own hands and, using Vercel’s v0, play with different implementations within an hour.
Thanks to AI, I can not only be a coder but also do some product discovery, for example playing with different mockups and designing a few alternatives. Even though the final results will still need some extra touches from a Product Designer, I’m much closer to building what’s right.
AI Won’t Replace Engineers—But It Will Replace Those Who Don’t Adapt
Marty Cagan says:
So if I’m right, over the next 3-10 years we’ll continue to see the average product team drop from something like 8 people (6 engineers, a product manager and a product designer), down to 3 (a product manager, a product designer, and an engineer).
That might be overly ambitious, but it points to a trend: If AI handles much of the coding, we need fewer pure coders. We still need strong engineers to validate AI outputs—debug, ensure scalability, confirm performance. But the emphasis shifts to how well you orchestrate the process, not how many lines you type.
The Future of Jobs Report by the World Economic Forum says:
The most common workforce response to these changes is expected to be upskilling workers, with 77% of employers planning to do so. However, 41% plan to reduce their workforce as AI automates certain tasks.
Today, in the early days of Gen-AI, we already know that creating source code can indeed be automated. And many people predict that the first ones to be replaced will be Junior Software Engineers.
I think it’s the opposite. Entry-level engineers who master the latest tools and ways of working will be able to outperform senior programmers who don’t.
Will AI replace software engineers? No—but software engineers using AI effectively will replace those who fixate on churning out lines of code instead of building the right products.
What Engineering Leaders Can Do
In this new era, your ability to discover the right solutions—paired with AI’s power to deliver them—creates a multiplier effect. Here’s how:
Invest in AI Fluency: Encourage your team to learn prompt engineering, code assessment, and advanced AI integration.
Focus on Problem Definition: Make sure everyone understands why a feature is being built. Data analytics, user research, and a strong product culture matter more than ever.
Empower Collective Problem-Solving: Good leaders ensure the whole team knows the product context, user journey, and success metrics, so they can do more than just “implement.”
There are no shortcuts to understanding your customers better, how your product is being used, and what real users’ problems you solve. But once coding becomes a commodity, these skills will be the differentiator for top in-demand engineers and engineering leaders.
Here are some of PEM’s past materials that can help you with that:
The 3-part series: The Role of Engineering in Product Model Transformation – How to shift from just software delivery to ensuring you build the right things.
Product Analytics for Engineering Leaders – Build your first product analytics dashboards so you don’t have to wait for product managers or product analysts to tell you how your product is being used.
The Problem Solving Framework – A framework to better understand the problem you’re trying to solve (agnostic of the technology used).
You Build It, You Run It – Software delivery is part of your work. Here’s the whole journey you should own.
The Bottom Line
AI will reshape software engineering, letting us accomplish more with fewer people. The best engineers won’t vanish—they’ll adapt to become orchestrators, strategists, and quality experts. Teams will be smaller, more focused on discovery, and more reliant on human insight than raw coding brute force.
When AI can code in seconds, your real value is deciding what’s worth coding in the first place.
Here’s Your Next Step:
Share this article with a colleague who’s still heads-down in “delivery” mode.
Comment Below about your biggest AI-related concern or excitement.
Subscribe to stay ahead of the AI wave and lead your team into a discovery-first future.
Thanks for reading!
Salute for a such a deep article