The Engineering of Intent, Chapter 36: The Long View

This is Part 36 of a series walking through my book The Engineering of Intent. In the previous chapter, we catalogued the archetypes. Chapter 36 steps back from practice entirely. What happens to our craft, to our profession, to our lives, over the next ten years?


I Will Not Pretend to Know. Here Are the Frames I Use.

The final chapter of Part IX doesn’t predict; it frames. Every major shift in our profession has followed a similar shape: early hype, plateau of disillusionment, slow integration into practice, a new normal that looks from the outside like it happened all at once but took a decade.

We are, today, somewhere between hype and integration. The prediction I am confident about: by 2030, the practices in this book will feel obvious. The controversies of 2026 will have settled; the frontier will have moved. The practices will have become assumed. The argument for them will no longer need making.


What Will Endure

Three things:

  • The discipline of clear intent. Whatever models we talk to, whatever tools we build, we will still need to say clearly what we want. The cost of ambiguous intent only grows as systems become more powerful.
  • The value of rigor under pressure. The temptation to cut corners will remain, and will remain seductive. The engineers who hold the line will remain the ones you want on your team.
  • The importance of craftsmanship. The systems that outlive their authors are the ones built with care. Care is expensive; care is worth it. In a decade, the companies that invested in craft will still be around.

What Will Change

  • Where the line between human and agent sits. Agents can execute within a codebase today; in the near future they’ll coordinate across codebases and services. The interesting work moves up the stack.
  • The cost structure of software. Software has been labor-intensive for fifty years; it will become more capital-intensive — compute, tooling, model licensing. Companies that don’t attend to compute unit economics will be surprised.
  • What “software engineer” means. The title will split. Deeper toward specialization (models, systems, security, scale) or broader toward product, design, strategy. The generalist middle thins out — agents cover it efficiently.
💡 The most practical prediction: The generalist-engineer role — the person who writes everything moderately well — will thin out over the next few years. If that’s you, pick a direction. Deeper or broader. Either works. The middle, historically safe, is the position that compresses most.

A Final Word on Identity

“If your identity is tied to the act of typing code, the next few years will be destabilizing. If your identity is tied to solving problems well, they will be invigorating. The shift is available to anyone willing to make it. It is not an easy shift — it involves letting go of something real that the industry rewarded for decades — but it is a shift with a long runway of growth on the other side.”

The sentence I’d put on the wall: You are not your editor. You are not your language. You are not even the craft of software, strictly speaking. You are a thoughtful person who makes useful things. That is a larger identity than any one tool or skill. It has survived every previous wave of change in this industry. It will survive this one.

Take care of yourself. Take care of each other. And keep building.


Next up — Chapter 37: Context Scaling. Part XXIII of the book contains four advanced-topic chapters that extend the core practice into harder territory. Chapter 37 opens the section with context scaling — just-in-time retrieval patterns for million-line codebases.


📖 Want the full picture?

The chapter walks the hype-integration cycle in full, the three things that will endure and the three that will change, and the identity reflection that readers tell me they return to on hard weeks.

Get The Engineering of Intent on Amazon →

2026-05-22

Sho Shimoda

I share and organize what I’ve learned and experienced.