You Don't Need to Become an Engineer. You Need to Become a Systems Thinker.

I see both pipelines in my inbox every week.

On one side: companies desperate for people who can design AI agent architectures. Orchestration layers, governed workflows, autonomous systems wired into real business logic. Compensation north of $400K. Relocation offered anywhere.

On the other side: experienced professionals with solid track records, suddenly competing with hundreds of applicants for roles that pay less than they earned two years ago.

Same industry. Same week. Two completely different realities.

I've been thinking about why the gap is so wide, and I keep landing on the same conclusion. The dividing line between these two groups has almost nothing to do with technical skill. It has everything to do with how people think about work.

I watched this exact pattern before. In the mid-90s, Sweden digitalized fast. Government offices running on paper binders and physical archives suddenly needed databases, email, digital workflows. Driver's licenses went electronic. Tax filings moved online. Every company started requiring "computer skills" in job listings.

The anxiety was enormous. People who had built entire careers on being organized and thorough felt obsolete overnight. The assumption was that you needed to become a computer nerd to survive. Managers signed up for evening courses on DOS commands. Everyone thought you needed to understand how the machines actually worked, down to the hardware.

Most of that fear was misplaced. The people who thrived weren't the ones who learned to program. They were the ones who understood that their job was fundamentally the same, but the tools had changed. The accountant who learned spreadsheets instead of ledger books. The project manager who coordinated through email instead of phone trees. They didn't become engineers. They became digitally fluent versions of what they already were.

That parallel is playing out again right now, but faster and with higher stakes.

The Upskilling Pattern

Every major technology transformation follows the same arc. A new capability arrives. It feels alien. People assume they need deep technical expertise to survive. Then reality settles in: what's actually required is that everyone moves up one level of abstraction.

Digitalization required understanding digital systems. Not building them. Using them, configuring them, thinking in data flows instead of paper trails.

The cloud wave did the same thing a decade later. You didn't need to become a distributed systems engineer. But architects had to think about availability zones and scaling patterns. Product managers had to understand deployment pipelines. The abstraction level went up for everyone, not just the infrastructure team.

AI is doing this again. From executing tasks to orchestrating systems that execute tasks. From writing the report to designing the workflow that produces, reviews, and refines the report. From being the person who does the work to being the person who designs how the work gets done.

The shift is harder to see when you're in the middle of it. People focus on learning specific tools, which model to use, how to write better prompts. Those things matter, but they're the equivalent of learning which buttons to click in Excel. The real transformation is in how you think about the work itself.

The New Skill Stack

I've spent the last year running agentic workflows daily. Multiple AI agents coordinating to handle classification, drafting, review, and escalation across real business processes. The experience has made something clear: the skills that matter most are not technical. They're thinking skills and people skills.

Systems thinking. Understanding how parts connect. Seeing feedback loops. Knowing that when you change one part of a workflow, three other parts shift in response. If you can look at a business process and see it as a system of interconnected components rather than a list of tasks, you're ahead of most people. This skill has been taught in business schools for decades. It has never been this relevant.

Orchestration. Not coding agents. Designing how they collaborate. What does each one handle? Where are the handoffs? What happens when one produces garbage and the next one has already consumed it? This is closer to project management than software engineering. The best orchestrators I've worked with think like producers, not programmers. They see the whole production and know when to intervene.

Communication with non-human entities. Specifying what you want from an AI system clearly enough that it actually delivers. Verifying the output. Calibrating over time. It's a new form of delegation, and it requires the same instincts as managing a junior team member: clarity, patience, and the ability to spot when something is 90% right but critically wrong in the last 10%.

Social skills and leadership. The more work agents handle, the more important human judgment becomes. Empathy, stakeholder management, trust-building across teams, the political skill to navigate organizational change. When the execution layer is automated, the human value concentrates in things that require genuine human presence. Leading through ambiguity. Making decisions with incomplete information. Reading a room.

Entrepreneurial drive. Driftighet, as we say in Swedish. Initiative. The willingness to try things without waiting for permission. Building a prototype over the weekend because you had an idea. Testing a new workflow on your own before proposing it to the team.

The pattern I see in the strongest candidates is consistent. They talk about failures, not demos. They lead with the business problem, not the technology. They've thought about what happens at 3 AM when nobody is watching and the system breaks. They have actual opinions about orchestration tradeoffs because they've hit the walls of their tools in practice.

Compare that to someone whose AI experience is "I use Copilot daily and I've built several GPT wrappers." That's real, it's relevant, and it's the AI equivalent of being proficient in Microsoft Office. It's not enough.

The Consultancy Advantage

There's an irony here that I find interesting. The skills I just described are exactly the skills that consulting firms have cultivated for decades. Structured thinking. Client communication. Stakeholder management. The ability to land in an unfamiliar organization, understand its problems quickly, and deliver under pressure. Resourcefulness. Driftighet.

The tech world has historically undervalued these skills. "Soft skills" gets used as a polite way of saying "not the real work." I've been in enough rooms to know this bias is real.

The AI era is inverting that hierarchy. The technical execution layer is increasingly handled by AI systems. What remains is the ability to think clearly about what needs to be built, communicate it precisely, lead people and agents through the process, and verify that the result solves the actual business problem. The "soft" skills are becoming the hardest requirements.

I saw this happen recently at a hackathon we organized. A managing director at a sister company joined for the first time. Something clicked. Within weeks she was using Claude Code to build Chrome extensions and spreadsheet scripts that automated parts of her daily workflow. She's not an engineer. She's a business leader who started thinking in systems. She saw her own work as a set of processes that could be designed, automated, and improved. That shift in perspective was worth more than any coding bootcamp.

Who Rises

I have a profile in my head of who thrives in this transition.

The person who hears about a new tool on Tuesday and has built something with it by Thursday. Not because someone told them to. Because they were curious.

The person who, when a process is broken, doesn't file a ticket and wait. They sketch a better version, test it, and present the result.

Charming, bold, entrepreneurial doers. People with initiative and the social skill to bring others along. Nate B. Jones called it "the $400K divide," and I think that framing is accurate. On one side: people designing AI systems, pulling $400K+ offers before you're halfway through your hiring process. On the other: experienced professionals watching the market compress around them. The gap widens every quarter.

The people who cross to the right side exist at every level of every organization. They're not always the loudest. But they're always the ones who move first.

Start Today

You don't need to become an engineer. You don't need to understand transformer architectures or fine-tune models.

You need to become a better systems thinker. Look at your work and ask: what's the system here? What are the inputs, outputs, feedback loops? Where would an AI agent add the most value?

Start small. Set up one workflow where an agent handles a task you currently do manually. See what breaks. Fix it. Iterate. The learning is in the doing.

And do it now. Adding "AI skills" to your résumé is the equivalent of adding "cloud experience preferred" to a job listing in 2014. It checks a box without changing anything fundamental.

The 90s didn't require everyone to become a programmer. They required everyone to think digitally. This era doesn't require everyone to become an AI engineer. It requires everyone to think in systems.

The people who start today will have a meaningful advantage over the ones who wait for someone to tell them what to do. That has always been the pattern. It hasn't changed.