You didn't get replaced because you weren't good
You got replaced because the math changed.
A junior copywriter cost $58,000 a year plus benefits, took two weeks to onboard, and produced a draft in three hours. A model that costs a fraction of a cent per call now produces that same draft in eleven seconds, doesn't need PTO, and ships at 3am. What changed was the unit economics of the task, not your talent at performing it.
You didn't get replaced because you weren't good. You got replaced because the math changed. Specifically: a class of cognitive tasks that used to require a salaried human can now be done in seconds for fractions of a cent, with no onboarding and no PTO.
This is lesson one of chapter zero of a thirty-chapter course on building with AI. It's an identity-repair lesson before it's anything else. We will not skip this part. The mid-career adults who don't read this lesson are the ones who quit ch01 in seven days, because they walked into the syntax believing the layoff was their fault and the syntax confirmed it.
The layoff was not your fault. Let's name what actually happened.
Five fields, real numbers
Between 2024 and 2026, the categories that got cut hardest were the exact ones where AI was best:
- Software engineers. Big-tech headcount peaked in mid-2022 and has fallen every quarter since. The juniors went first — entry-level SWE hiring at FAANG companies roughly halved (50%+) from 2022 to 2026. Copilot and Claude Code didn't replace senior engineers. They replaced the pipeline that became senior engineers.
- Copywriters and content marketers. Agencies that ran 40-person copy floors in 2022 now run 6-person ones. The work didn't shrink — the agencies still ship the same volume — but four people with Claude produce what forty people used to.
- Customer service reps. Tier-1 support — the people who read scripts and routed tickets — got hit first and hardest. Klarna let attrition shrink headcount ~40% while saying an AI assistant did the workload-equivalent of 700 human agents — then began rehiring contractors in 2025 after quality complaints. Every CS-heavy SaaS company followed.
- Graphic designers and junior creatives. Stock-asset designers, social-media designers, junior product designers. Midjourney and Figma's AI tools made the "I need 200 banner variants by Friday" workflow a 90-second job.
- Paralegals and junior analysts. Doc review, contract summarization, citation checking, first-pass research memos. Anything that was "read this 400-page document and tell me what's in it" is now $0.40 worth of model inference.
Five fields. Different headlines, same underlying mechanic. The model got faster at the parts of those jobs that were structured, repetitive, and high-volume. The math changed.
What "the math changed" actually means
It does not mean you were doing the wrong work. It does not mean your field was fake. It does not mean the years you spent getting good were wasted. It means a tool got built that is cheaper than you at the parts of your job that scale.
This is not the first time this has happened. The same math change happened to typesetters in the 1980s, to travel agents in the late 1990s, to film photographers in 2005, to taxi dispatchers in 2012, to mall retail in 2018. In every case, the people who landed best were the ones who:
- Named the math change quickly. Spent zero time arguing that it shouldn't have happened.
- Inventoried what survived. The parts of the job the new tool couldn't do — and aimed for those.
- Got fluent in the new tool fast. Not as users. As builders or operators.
That third step is what this course is for. The first two are what this lesson is for.
"You're not behind. You're early."
If you're reading this lesson because you got cut in 2025 or 2026, here's the honest read of where you are.
If you're here because you got cut, you're not behind. You're early.
Most of the workers who will be cut in the next eighteen months haven't been cut yet. The middle-manager layer in big-co tech has barely started. Mid-tier law firm associates haven't started. Insurance underwriters haven't started. Pharma medical writers haven't started. A second wave is coming, and the people who got cut in the first wave are the ones who get the eighteen-month head start on what comes next.
You're not late to AI. You're early to "people who used to do X for a living, and now build the tools that do X."
What this chapter does, plainly
Chapter 00 has four lessons. This one is identity repair. The next three are mental-model installation:
- Lesson 02 explains what an LLM actually is, so the rest of the course stops being magic.
- Lesson 03 explains why your old job already taught you 80% of how to talk to one.
- Lesson 04 maps where this course goes and which path through it matches where you came from.
By the end of this lesson — in about sixteen minutes — you will have typed your first line of Python. It will be one line. The point is not the Python. The point is the moment you go from "person who got cut" to "person who started building." That moment is small on purpose. We don't need it to be dramatic. We just need it to happen once, so the rest of the course has somewhere to start.
You didn't get replaced because you weren't good
You got replaced because the math changed.
A junior copywriter cost $58,000 a year plus benefits, took two weeks to onboard, and produced a draft in three hours. A model that costs a fraction of a cent per call now produces that same draft in eleven seconds, doesn't need PTO, and ships at 3am. What changed was the unit economics of the task, not your talent at performing it.
You didn't get replaced because you weren't good. You got replaced because the math changed. Specifically: a class of cognitive tasks that used to require a salaried human can now be done in seconds for fractions of a cent, with no onboarding and no PTO.
This is lesson one of chapter zero of a thirty-chapter course on building with AI. It's an identity-repair lesson before it's anything else. We will not skip this part. The mid-career adults who don't read this lesson are the ones who quit ch01 in seven days, because they walked into the syntax believing the layoff was their fault and the syntax confirmed it.
The layoff was not your fault. Let's name what actually happened.
Five fields, real numbers
Between 2024 and 2026, the categories that got cut hardest were the exact ones where AI was best:
- Software engineers. Big-tech headcount peaked in mid-2022 and has fallen every quarter since. The juniors went first — entry-level SWE hiring at FAANG companies roughly halved (50%+) from 2022 to 2026. Copilot and Claude Code didn't replace senior engineers. They replaced the pipeline that became senior engineers.
- Copywriters and content marketers. Agencies that ran 40-person copy floors in 2022 now run 6-person ones. The work didn't shrink — the agencies still ship the same volume — but four people with Claude produce what forty people used to.
- Customer service reps. Tier-1 support — the people who read scripts and routed tickets — got hit first and hardest. Klarna let attrition shrink headcount ~40% while saying an AI assistant did the workload-equivalent of 700 human agents — then began rehiring contractors in 2025 after quality complaints. Every CS-heavy SaaS company followed.
- Graphic designers and junior creatives. Stock-asset designers, social-media designers, junior product designers. Midjourney and Figma's AI tools made the "I need 200 banner variants by Friday" workflow a 90-second job.
- Paralegals and junior analysts. Doc review, contract summarization, citation checking, first-pass research memos. Anything that was "read this 400-page document and tell me what's in it" is now $0.40 worth of model inference.
Five fields. Different headlines, same underlying mechanic. The model got faster at the parts of those jobs that were structured, repetitive, and high-volume. The math changed.
What "the math changed" actually means
It does not mean you were doing the wrong work. It does not mean your field was fake. It does not mean the years you spent getting good were wasted. It means a tool got built that is cheaper than you at the parts of your job that scale.
This is not the first time this has happened. The same math change happened to typesetters in the 1980s, to travel agents in the late 1990s, to film photographers in 2005, to taxi dispatchers in 2012, to mall retail in 2018. In every case, the people who landed best were the ones who:
- Named the math change quickly. Spent zero time arguing that it shouldn't have happened.
- Inventoried what survived. The parts of the job the new tool couldn't do — and aimed for those.
- Got fluent in the new tool fast. Not as users. As builders or operators.
That third step is what this course is for. The first two are what this lesson is for.
"You're not behind. You're early."
If you're reading this lesson because you got cut in 2025 or 2026, here's the honest read of where you are.
If you're here because you got cut, you're not behind. You're early.
Most of the workers who will be cut in the next eighteen months haven't been cut yet. The middle-manager layer in big-co tech has barely started. Mid-tier law firm associates haven't started. Insurance underwriters haven't started. Pharma medical writers haven't started. A second wave is coming, and the people who got cut in the first wave are the ones who get the eighteen-month head start on what comes next.
You're not late to AI. You're early to "people who used to do X for a living, and now build the tools that do X."
What this chapter does, plainly
Chapter 00 has four lessons. This one is identity repair. The next three are mental-model installation:
- Lesson 02 explains what an LLM actually is, so the rest of the course stops being magic.
- Lesson 03 explains why your old job already taught you 80% of how to talk to one.
- Lesson 04 maps where this course goes and which path through it matches where you came from.
By the end of this lesson — in about sixteen minutes — you will have typed your first line of Python. It will be one line. The point is not the Python. The point is the moment you go from "person who got cut" to "person who started building." That moment is small on purpose. We don't need it to be dramatic. We just need it to happen once, so the rest of the course has somewhere to start.