Everyone’s Watching the Wrong Movie
The discourse around “AI replacing developers” has become exhausting. We’re all arguing about the wrong thing.
One camp is screaming: “AI will take all our jobs! Learn to prompt or perish!”
The other camp is laughing: “AI can’t even handle production-grade distributed systems! We’re safe!”
Meanwhile, I’m over here watching something completely different happen. And it’s way scarier than either narrative.
The market doesn’t need as many of us anymore.
Not because AI got “good enough” to replace us. But because the gap we were filling is closing from a completely different direction.
Let Me Tell You About Bob
Bob runs a small development shop. Makes decent money building apps for local businesses. Nothing fancy – inventory systems, booking platforms, internal dashboards. Bread and butter work.
Karen owns a small restaurant chain. She has accounting problems. Her invoices are scattered across folders, formats are inconsistent, reconciliation is a nightmare.
The Old World:
- Karen realizes: “I need software for this”
- Karen emails Bob: “Can you build me an invoice reconciliation tool?”
- Bob quotes $3,000, two-week timeline
- Karen either pays or continues suffering manually
- Bob feeds his family
The New World:
- Karen has the same problem
- Karen opens ChatGPT: “I need to compare these Excel files and show me which rows are different”
- ChatGPT spits out a Python script
- Karen runs it on her laptop
- Problem solved
- Bob never gets the email
“But wait,” the engineers cry, “what happens when it breaks? Surely she’ll need Bob to maintain it!”
You think like a programmer. Go touch some grass in the Accounting department.
When Karen’s Excel pivot table breaks—when the formula throws a #REF! error or the VLOOKUP breaks—does she call Bob the Excel Consultant? Does she open a Jira ticket? Does she enter a 3-week maintenance cycle?
No. She mutters “ugh this stupid thing,” hits Ctrl-A, Delete, and spends 12 minutes making a new one from scratch. She treats it like a sticky note, not a cathedral.
Cost of regeneration: $0.
Cost of her time: ignored.
Call to Bob: never happening.
The script broke? She’ll just prompt a new one. NewInvoiceBot_v7.py replaces NewInvoiceBot_v6.py the same way she made a new Excel sheet for Q4 instead of fixing the Q3 one. No maintenance culture exists because no asset value was assigned.
You were maintaining code because you viewed it as infrastructure. Karen views it as consumable.
The Localhost:3000 Revelation
You know what the big dogs love to mock? Vibe coders sharing screenshots of their localhost:3000 projects.
“Lmao you didn’t even deploy it? That’s not real software!”
But here’s the thing they’re missing: for most people, localhost IS the deployment.
The localhost:3000 screenshot isn’t amateur hour. It’s the Excel-fication of code. Disposable. Ephemeral. Infinitely regeneratable.
We laughed BUT we missed the bigger precedent: Excel already proved that business users prefer disposable regeneration over maintenance.
Karen doesn’t need “software as a service.” She needs “software as a consumable.” Like a paper towel. Use it, crumple it, grab another.
The Infrastructure We Built is Post-Tragedy Necessity
Here’s where it gets existential.
Why do we have:
- Authentication systems?
- Multi-tenancy?
- Cloud infrastructure?
- PCI-DSS compliance?
- Load balancers?
- API gateways?
- Maintenance retainers?
Not because they’re inherently necessary to solve problems.
They’re necessary because we built an entire industry around SELLING software as assets to people who couldn’t make it themselves. We convinced businesses that software is capital expenditure—something to maintain, depreciate, and protect.
But go look at Karen’s desktop. How many “broken” Excel files are sitting there? Zero. She deleted them. How many “deprecated” Python scripts? She doesn’t keep deprecated things because there’s no cost to regeneration.
How Many of You Actually Work at Netflix Scale?
Every AI-replacement thread devolves into:
- “AI can’t handle distributed systems at scale!”
- “What about microservices architecture?”
- “Complex algorithmic optimization!”
Cool. How many developers actually work on those problems?
Maybe 5%?
The other 95% are:
- Building CRUD apps (Excel with extra steps)
- Making forms talk to databases (Excel with a web UI)
- Integrating third-party APIs (Excel VLOOKUP to external data)
- Building internal dashboards (Excel charts, but slower)
- WordPress sites (Excel for websites)
- E-commerce platforms (Excel with checkout)
- Simple mobile apps (Excel on your phone)
This is EXACTLY the work AI is getting scary good at. And it’s exactly the work that can be treated as consumable rather than maintained.
The Disposability Gap is Everything
When Karen can solve 70% of her own problems with disposable, regeneratable tools, she doesn’t accumulate technical debt. She throw away the debt with the tool. Bob’s business just shrank by 70%, but he’s still waiting for the maintenance call that’ll never come because there’s nothing to maintain—only things to regenerate.
Will people still build software? Yes.
Will “Software Engineer” be a large, distinct profession in 10 years? Maybe not. Or rather, it’ll be split between:
- Complexity Surgeons (the 5% doing distributed systems, security, stateful multi-user nightmares)
- Literate Business Users (the 95% who “write code” the way they “write Excel formulas”—disposably, daily, without calling it “engineering”)
What I AM Saying
Getting replaced is scary but clean. It’s dramatic. You see it coming.
Irrelevance is insidious.
It’s Bob watching his client calls drop from 20/month to 15 to 10 to 5, each time thinking “just a slow quarter,” not realizing that Karen now treats code like she treats spreadsheet cells—disposable, regeneratable, not worth a phone call.
It’s looking at your Kubernetes certification and realizing the market is moving toward infinite sticky notes that need no orchestration because they’re used once and deleted.
It’s defending against the headshot while bleeding out from deprofessionalization.
Look outside your cubicle. Look at Accounting. Look at Marketing. They don’t maintain their Excel messes—they remake them. And they’re about to treat your beautiful REST APIs and React components the exact same way.
PS: Yes, I wrote this with AI assistance. Because I’m not a Lotus 1-2-3 consultant pretending Excel doesn’t exist. I’m adapting.
PPS: If this made you angry, hit that “Sign Out” button. But first, go ask Karen in Accounting how many times she “maintained” an Excel formula versus just making a new sheet. I’ll wait.
PPPS: If this resonated, you should probably start building your ark. The flood isn’t coming—it’s already here. We’re just standing in ankle-deep water arguing about whether Kubernetes can save us while Karen is throwing away Python scripts like used Kleenex.