Real People, Real Jobs, Replaced by AI: Stories From 2025-2026

CONTENT MARKETING

"Monday morning, our Slack was disabled. By Friday, 4 of 5 of us were gone."

Sarah (not her real name) was a mid-level content writer at a SaaS company in Austin. Her team of five writers produced blog posts, case studies, and email newsletters. In March 2025, the company announced they were "restructuring the content department around AI-first workflows."

The result: four writers were laid off. One senior writer was kept — retitled as "AI Content Director" — and given ChatGPT Enterprise, Jasper, and a mandate to produce the same output volume the five-person team previously handled. Within two months, output actually increased.

"They didn't even pretend it wasn't about AI," Sarah told me. "The CTO said in the all-hands: 'We're not replacing all content roles, we're replacing content production with AI-assisted production.' I was a 'content production' role. The senior writer who stayed was the one who'd been pitching AI integration for six months."

The lesson: The writer who positioned herself as an AI operator survived. The writers who just wrote well didn't. Quality of writing had nothing to do with who stayed and who left.

Source: Interview with affected employee, corroborated by LinkedIn layoff posts and company Glassdoor reviews, March 2025.

CUSTOMER SUPPORT

"The AI handles 80% of tickets now. We went from 50 agents to 18 in six months."

Klarna made headlines in 2024 when their CEO announced their AI assistant handled the work of 700 full-time agents. But it's not just fintech giants. A mid-sized e-commerce company in the Midwest (I agreed not to name them) deployed an LLM-powered support system in Q3 2025.

Before AI: 50 customer support agents handling ~8,000 tickets per week. Average handle time: 12 minutes. Customer satisfaction: 82%.

After AI: 18 agents handling the 20% of tickets the AI escalates (~1,600/week). Average handle time for those complex tickets: 15 minutes. Customer satisfaction: 79%. Cost reduction: roughly $1.3 million annually.

"The 32 people who got laid off weren't bad at their jobs," a former team lead told me. "They handled password resets and order tracking. The AI does that instantly now. The 18 who stayed handle angry customers, complex refunds, and edge cases the AI can't figure out. The job changed from 'answer lots of tickets' to 'handle the hardest tickets.' Totally different skill set."

The lesson: Volume-based support work — handling many simple, similar requests — is exactly what AI excels at. The survivors were the ones who could handle the messy, emotional, non-standard cases.

Source: Industry contacts, corroborated by Gartner customer service automation reports and Klarna's public disclosures.

SOFTWARE ENGINEERING

"We used to hire 15-20 junior devs a year. In 2025, we hired 3. And we're not unique."

This one's less about layoffs and more about a hiring freeze that became permanent. A senior engineering manager at a publicly traded tech company (think: household name, 5,000+ employees) walked me through their hiring data.

2023: 18 junior engineers hired. Average ramp time to productivity: 4-6 months.
2024: 12 junior engineers hired. AI coding tools deployed company-wide. Senior engineer productivity up ~40%.
2025: 3 junior engineers hired. Explicit strategy shift: "fewer, more senior engineers with AI amplification."
2026 (projected): 0-2 junior hires.

"It's not that we don't need engineering work done," he explained. "It's that one senior engineer with Cursor and Copilot produces more than a senior plus two juniors used to. The junior role was our talent pipeline — that's how people became seniors. We're eating our seed corn and we know it, but the short-term productivity gain is too big to ignore."

The people affected aren't laid-off employees — they're the 2024 and 2025 CS graduates who can't find entry-level roles. Coding bootcamp placement rates have dropped from ~80% to under 40%, according to industry reports.

The lesson: The "learn to code" advice that dominated the 2010s assumed an infinite appetite for entry-level developers. AI broke that assumption. The on-ramp to a software career is narrower than it's been in 15 years.

Source: Interview with engineering manager, corroborated by industry hiring data, bootcamp placement reports, and tech company earnings calls discussing AI productivity gains.

TRANSLATION & LOCALIZATION

"Clients don't call anymore. They run it through DeepL or ChatGPT and call it done."

Miguel ran a small translation agency in Miami for 12 years, employing 22 translators handling English-Spanish-Portuguese business and legal documents. Revenue peaked at $2.1 million in 2022.

By late 2025, revenue was $640K. He's down to 7 translators, and most of their work isn't translation anymore — it's "post-editing" machine-translated text. A client sends AI-translated documents, and his team reviews them for legal accuracy and nuance.

"The machine translation got too good, too fast," Miguel says. "In 2022, AI translations were awkward — you could tell. By 2025, for business documents? Maybe one error every 3-4 pages. Most clients don't need human translation anymore. They need a human to verify the AI didn't mess up the legal clause on page 7."

The 15 translators he let go weren't bad translators. They were specialists in document types that AI now handles well (business correspondence, simple contracts, marketing brochures). The 7 who stayed are legal translation specialists who handle complex contracts where a single mistranslated word could cause a lawsuit.

The lesson: When AI becomes "good enough" for a task, the human role shifts from producer to verifier. The people who survived were the ones whose work had consequences that were too severe to trust to AI alone.

Source: Interview with agency owner, corroborated by translation industry revenue data and ATA (American Translators Association) market reports.

GRAPHIC DESIGN

"I graduated top of my class. Six months later, I'm working retail. AI did my dream job better and faster — for free."

Jordan graduated from a well-regarded design program in May 2025 with a strong portfolio. Applied to 80+ junior designer positions. Got 4 interviews. Zero offers.

"Every interview was the same," Jordan told me. "They'd ask if I use Midjourney or Firefly. I'd say I can, but my real skill is original design work. And they'd get this look — like I'd just told them I ride a horse to work."

One creative director was blunt: "We used to need 3 juniors to produce the volume of social graphics, banner ads, and landing page illustrations our marketing team needs. Now I have one junior who's amazing at Midjourney prompts, and she produces more than the old team of 3. I feel bad about it, but the economics are what they are."

Jordan's portfolio was better than the AI-generated work. But it took Jordan 3 days to produce what the AI + a prompt-skilled designer could produce in 3 hours. In a volume-driven role like junior design, speed matters more than craft.

The lesson: When the output quality gap is small but the speed gap is enormous, businesses choose speed. Junior designers need to be AI-native, not AI-adjacent. The "I can use AI if I need to" attitude isn't enough — you need to be better at AI-assisted design than the people who've been doing it for 2 years.

Source: Interview with recent graduate, corroborated by design industry hiring trends, AIGA surveys, and job board data.