Software engineering isn’t dying. It’s having its fast fashion moment.

Everyone sees AI writing code and declares the profession over. Wrong takeaway.

The real shift is simpler and more familiar: software is moving from artisan craft to industrialized production with distinct market tiers. This has happened before. The textile industry spent 100 years making this exact transition.

Textiles, The Canonical Example

Before industrialization, cloth was made by skilled artisans working by hand. Mechanization arrived in waves — the flying shuttle, the water frame, the power loom, steam power — embedding expertise into machines and collapsing production costs.

The profession didn’t vanish. The market reorganized.

Clothing did not converge to a single price or quality level. It split into tiers with different economics and different reasons to exist.

Where the Market Landed

Modern textiles operate across three durable tiers:

• Luxury and bespoke (~35–40% operating margins). Brands sell scarcity, craftsmanship, and status. Pricing power comes from trust, reputation, and perceived uniqueness.
• Durable, high-quality goods (~20–25% operating margins). Products justify premiums through reliability and longevity. Customers stay because replacement is costly or risky.
• Fast fashion and mass market (~5–10% operating margins). Products are inexpensive, replaceable, and optimized for scale. Profitability depends on volume and operational efficiency.

All three benefit from modern manufacturing and show similar gross margins. What differs is why and how much customers pay.

The Margin Trap

The gap between luxury margins and mass-market margins isn’t production cost — it’s pricing power. Luxury brands set prices. Volume brands take them.

AI is collapsing production costs across software in the same way mechanization did for textiles. Faster production does not determine pricing power.

• If you operate in a premium segment, AI expands margins.
• If you operate in a volume segment, AI enables survival.
• If your cost structure assumes premium economics in a volume market, AI accelerates failure.

Software is deep into its own industrialization cycle. Early systems were bespoke. Enterprise software scaled development through teams and processes. AI and low-code tools now embed increasing amounts of judgment directly into the tools.

Three tiers of software are forming that echo the textile tiers:

• "Luxury software" (35%+ operating margins). Buyers pay for quality - reliability, compliance, accountability, and deep integration. Quickbooks, Paychex, Visa are all examples in this space.
• Durable, high-quality platforms (~20–30% operating margins). Systems that become embedded in operations over time. Value comes from accumulated data, workflows, and switching costs. Faire, Zoom, Shopify, ServiceNow, Salesforce all live here.
• Fast-build and disposable software (~5–15% operating margins). Tools that solve immediate problems cheaply and can be replaced easily. Longevity is not the goal. These were solved by companies like Zapier, Canva, Bubble and are already being better solved on the back of Claude Code and personal software. Disruption is real and present in this tier.

AI does not collapse these tiers. It clarifies them.

The Tinkerers

But what about OpenClaw? Won't the AIs enable us to create any software at any time?

Yes, AI assistants, Claude code artifacts and so much more operate like the sewing machine did for textiles — they enable self-production without creating mass markets. Most of this work solves narrow, local problems and never becomes a product.

And, this personal software will grow rapidly because the barriers are low. Will it steal some of the market? Surely - largely by pushing out firms that didn't deserve to be there anyway. But it doesn't solve the same problems that well run professional software does - reliability, scale, and support still matter.

So what does this mean for you?

Implication #1 Talent is Changing, you need to evolve how you think about jobs

Welcome to the Engineer–Operator Split.

Industrialization changes labor as well as products.

Software long blurred roles because building and operating technical systems required the same scarce skills (coding) and the massive gross margins of the SaaS era enabled it-- that is no longer true.

Two distinct categories are emerging:

• Software engineers design systems where novelty is high or failure is unacceptable. Their work has long time horizons and architectural consequences.
• Builder-operators assemble, customize, and operate existing systems to solve concrete problems. They build on top of platforms rather than inventing new ones.

Mature engineering fields have always made this distinction. Software is catching up.

Mismatch is dangerous. Paying premium talent for commodity work destroys margins. Assigning commodity talent to mission-critical systems creates reliability risk. AI amplifies both outcomes by lowering the cost of doing the wrong thing faster.

The industry has done ourselves a disservice here by naming everyone who works with code a software engineer - that will need to be unwound (as will the everyone is a Product Designer or Product Manager versions).

Implication #2: Your organization needs to be clear on, and align what product you are selling to how you operate.

Cost pressures do not destroy companies on their own. Misalignment does.

In software, that means:

• Bespoke products built for customers unwilling to pay bespoke prices
• Commodity tools priced like enterprise platforms
• High overhead competing in volume markets
• Low-reliability products sold into high-stakes environments

AI didn't create these mismatches, but it is quickly exposing them.

The key question to ask yourself is "What Game Are You Playing?"

Every software business lives in a market tier, whether it realizes it or not. If your product disappeared tomorrow:

• Would customers rebuild it at any cost? You’re luxury.
• Would they migrate painfully but inevitably? You’re durable infrastructure.
• Would they switch in a weekend? You’re fast fashion.
• Would they stop doing the task entirely? You were never a product.

Your pricing, talent model, and strategy must match that reality. Most failures come from operating as if you occupy a higher tier than your customers believe you do.

AI won’t change which tier you’re in. It will remove the illusion that you’re in a better one. Software engineering isn’t dying. The market is revealing what kind of software actually gets to exist.