No one knows exactly where AI will take us beyond 12-18 months from now. Anyone who claims otherwise may have a bridge to sell you. But it’s clear that fundamental shifts in software engineering will happen within that time, and that they will transform every industry that runs on software in the process.
I’ve led software engineering at big tech companies like Microsoft, Snap and Google over the span of two decades. AI’s always been in the conversation and the innovation lab, but the changes happening now are truly unprecedented in both speed and scale. We have the opportunity to achieve more in the next year than we have in the last 10. By 2028, the digital economy we know today will look completely different. If it doesn’t, we’ve failed.
Despite headlines declaring the software profession dead, I’m confident its future is brighter than ever.
The Job Was Never About Typing
Many people who believe AI marks the end of software engineering misunderstand what the role actually is. The job has never been about writing code. It has always been about solving problems—and doing it in a way that reduces complexity, minimizes maintenance burden, and delivers something useful to the person on the other end. Code was the medium, not the purpose.
That distinction matters now more than ever. AI agents can write code. They can generate tests, scaffold services, wire up APIs, and produce boilerplate at a pace no human can match. What they cannot do—at least not yet—is decide what to build, understand why it matters, or navigate the tradeoffs that determine whether a system survives contact with the real world. That remains the engineer’s job. It always has been.
The era of the specialist coder—fluent in one language, deeply embedded in one stack—is giving way to the generalist orchestrator. Engineers will increasingly supervise fleets of agents that write business logic, generate tests, analyze logs, and suggest architectural changes. The daily work will center on mapping constraints, aligning outputs with product goals, and ensuring resilience and security.
Coding is no longer the special sauce. Context is.
Why Deep Understanding Still Matters
This shift to orchestration might sound like it lowers the bar. It doesn’t. If anything, it raises it.
To manage agents effectively—especially at scale, across large codebases and complex systems—you need to understand the underlying technology deeply. You need to know what good architecture looks like, how systems fail, where performance bottlenecks emerge, and when an agent’s output is subtly wrong in ways that won’t surface until production. An orchestrator who doesn’t understand the work being orchestrated isn’t an orchestrator. They’re a liability.
The essence of the job is shifting: from writing code to developing a deep understanding of how systems work, staying curious about technical details, and focusing relentlessly on building things that meet real needs. Systems-level thinking, lived experience, and nuanced judgment applied to agent output—that’s the new job description.
The Hazards No One Is Talking About
The benefits of agentic engineering are real. But there are three emerging hazards the industry needs to confront honestly, even setting aside the larger uncertainties around artificial general intelligence and societal disruption:
- Replenishment. If agents absorb the work traditionally handled by junior engineers—writing simple features, fixing bugs, learning a codebase—then we risk eliminating the profession’s apprenticeship layer. Junior roles have always been the on-ramp. Remove them, and the pipeline eventually dries up. The industry needs new models to replace that training ground, whether through structured apprenticeships, AI-assisted internships, or systems that teach engineers to work alongside agents from day one.
- Atrophy. I’ve spoken with dozens of engineers using AI agents extensively, and many describe a sense of skill erosion. When the agent handles implementation, you stop building the same intuition and muscle memory. Some say it’s harder to enter a flow state, the mental mode where many breakthroughs happen. This isn’t nostalgia. Engineers still need deep technical understanding to evaluate whether an agent’s output is correct, scalable, or safe. Otherwise, you’re reviewing code you can’t fully judge.
- Exhaustion. One surprise is how mentally draining agent management can be. Engineers get far more done, but many also report greater cognitive fatigue. Constantly switching between agent sessions, reviewing parallel workstreams, and maintaining coherence across semiautonomous systems creates a new kind of exhaustion. It’s productive work, but it’s intense in ways traditional engineering often wasn’t.
These aren’t theoretical concerns. They’re already happening, and they’ll become more pronounced as adoption accelerates.
The Economics: Jevons Paradox and the Cost-Cutting Trap
When steam engines became more efficient, people didn’t use less coal. They used more. As efficiency improved, costs fell, new applications emerged, and total demand surged. Economists call this the Jevons paradox, and it applies directly to AI and software engineering.
As AI makes engineers more productive, the cost of building software drops. But demand for software won’t stay flat. It will explode. Companies that once couldn’t afford custom tools will build them. Features that sat on the backlog because of limited engineering bandwidth will finally ship. Problems that weren’t worth solving suddenly become economically viable.
In the short term, the transition will be uneven, and for some engineers, painful. But the long-term direction is clear: Companies will need more engineers, not fewer.
In every case I’ve seen, companies that view AI primarily as a way to cut costs were already struggling businesses. AI is accessible to everyone, which means the market will favor greater and faster innovation, not simply the same level of innovation but cheaper. The companies cutting headcount today will be overrun by the companies that are expanding while migrating their entire teams to work effectively with AI. It’s like a tug of war where one side keeps adding stronger players. The other side’s only option is to match them, not drop the rope and declare they’ve saved on labor costs.
Beyond Software
AI won’t just change the software engineer’s daily workflow. It will expand the boundaries of what engineers can do.
When the cost of implementation drops dramatically, engineers can think more broadly. Problems that once required months of dedicated coding become afternoon projects. And the scope of what an engineer can take on expands beyond software into hardware, into cross-domain systems, into problems that were previously the exclusive territory of other disciplines. An engineer who can orchestrate AI agents effectively isn’t limited to writing web services or mobile apps. They can prototype physical systems, model complex processes, and solve problems in industries they’ve never worked in before.
This is the part of the future that excites me most. The role of the engineer is expanding, not contracting.
It’s Fun
All the risks and uncertainties aside, I want to say something that gets lost in the hand-wringing: Agentic engineering is fun. It is genuinely, deeply enjoyable.
Engineers are not leaning into this technology because they have to. They’re leaning in because building things has always been the point, and AI lets them build more, faster, at a scale that wasn’t previously possible. The creative leverage you get from working with a capable agent—spinning up an idea, iterating on it in real time, watching something come together in hours instead of weeks—is intoxicating. It’s the reason most of us got into this field in the first place.
Where This Leads
We are in the early stages of a transformation larger than the iPhone, the internet, and the desktop computer combined. The future of work isn’t humans versus machines. It’s humans working alongside AI agents—across every role, every industry, every level of an organization.
The engineering profession will look different. The daily work will change. The skills that matter will shift. But the core of the job—solving hard problems, reducing complexity, building things that work—isn’t going anywhere. If anything, it’s about to matter more than it ever has.
We’ve always been limited by our capacity to solve problems, not our ability to think of them. AI is about to dramatically expand that capacity. The question isn’t whether we’ll need engineers. It’s whether we’ll have enough.
