The world changed. Not “is changing,” changed. Past tense, the inflection in late 2025.
AI will write 99% of the code. It will do 99% of the debugging, deployment, and configuration. It will do the heavy lifting of code review. It will handle 95% of testing. You can argue with the numbers. You can’t argue with the direction. If you’re a software engineer and this doesn’t land with a thud in your gut, you’re either way ahead of the curve or in denial. There’s not much in between.
Software engineers are now like every other engineering discipline. Our primary job is no longer to manufacture software. It’s to architect, design systems, and scale through automation. We must develop an acute sense of what “great” looks like and become experts at delivering precise feedback. The craft isn’t gone; craft is still craft.
The same is true for designers. A software designer’s primary job was never really to draw pixels. It’s taste. It’s deep understanding of human interaction, comprehension, and emotions. It’s building design systems that scale and express a brand, a voice, and a feeling. AI will enable designers to focus on design and not proficiency with the tools du jour.
And Product Managers? Their job is as important as ever, maybe more so. In a world where it’s 100x easier to build anything, deciding WHAT to build, WHY to build it, and WHO to build it for is paramount. The PM’s job is to build a business through the product. Strategy, marketing, storytelling, tactical evolution of the product portfolio, all rooted deeply in an obsession to understand and serve the customer. When you can build anything, knowing what not to build becomes your superpower.
Roles will consolidate. Teams will be smaller, tighter, more nimble. But everyone, everyone, must develop taste and judgment. Everyone must become expert at delivering precise, actionable feedback. Velocity, acceleration, evolution: these are everyone’s job now. It’s a competitive imperative. And advantages compound. The question isn’t whether to change. It’s how fast you can accelerate.
Let’s talk about what’s actually happening on the ground. We’re entering a period of intense identity crisis. Software developers are mourning their love of coding, that feeling of flow, of craftsmanship, of solving a puzzle with your own hands. Some are sprinting ahead. Some will walk at a measured pace. Others will drag their feet in denial, clinging to a world that’s already gone.
Recently one of our engineers shared something that really stuck with me:
“I still feel this sadness about what I’m making not really having value anymore. I hope this will pass with time and practice. Where your effort used to go into specific functions being nice and maintainable, you now basically wouldn’t look at the code anymore, and instead focus on specifying things better… Having to let go is tough and makes you confront your identity. I think this is why there is often such vitriol and pushback to AI. A lot of my self worth and esteem is made of skills built over time with a lot of effort. Letting go of that forces you to reimagine your value.”
Read that again. Letting go of that forces you to reimagine your value. That’s the real work here. Not learning prompts. Not picking the right tools. Reimagining your value.
The good news? There’s a path through. That same engineer continued:
“I allowed myself a week of feeling depressed and lost about it and now coming out of the other side of it I’m feeling a lot better. The AI is still quite dumb: It doesn’t know anything about what it should be doing. It forces you to enter a new skill focus, which is specifying things very clearly and having a really testable approach to code… if you want the robot to do your work for you, you have to specify.”
The craft isn’t dead. It’s different. Forging outcomes is the craft now. Clarity of thought, precision of intent, obsessive attention to “what does good look like”: these were always the hard parts, especially in teams! Now this is the craft.
Some will find the pace of progress addicting. Others will revel in their ability to precisely specify intent and see it manifested. And some will be sad for a while, and that’s okay too. Meet people where they are. Let them grieve. Then help them find their new identity through practice.
So what should leaders do?
Invest NOW. Budget 10% of a builder’s salary on AI tooling. This is not the time to be stingy. At the bare minimum, budget $500/month per builder. There is no single additional “hire” that will yield the same ROI to your team. Redirecting funds for a single open headcount would even go a long way.
Identify your sprinters. Find the people who are already leaning in, already experimenting, already building with these tools. Charge them with leading from the front. There’s no substitute for learning by doing. The whole industry is simultaneously figuring this out, and it is an exciting time to be building.
Rethink team structure. How you form teams, their scope, how they collaborate, their size and composition: all of it needs fresh eyes. Consider Conway’s Law: your system architecture will mirror your org structure. If you want nimble, adaptive systems, you need nimble, adaptive teams.
Hire humans. Remember that teams are made of individuals. Maximize the combination of individual strengths for maximum collective value. Hire for diversity across all dimensions. Maximize human uniqueness, creativity, and passion. AI can do the mechanical work. It cannot do the human work.
Focus intensely on WHAT to build and WHY. Product judgment is your moat now. Chart a more ambitious roadmap. Make bigger, bolder bets.
Accelerate. Not recklessly, deliberately. Advantages compound. The companies that figure this out first will be pulling away while others are still deciding whether to start.
Identify your sprinters. Invest in them. Let them lead.
The world changed. Accelerate.