TB: Your Job Isn't Disappearing — It's Shrinking Around You in Real Time (Tegze)
Core Thesis
Jan Tegze argues the popular AI-jobs frame is wrong twice: jobs are not disappearing wholesale, and the people pretending it is "business as usual" are losing ground in real time. The actual dynamic is shrinkage — your tasks compress, your hourly value drops, and the only real defense is to redesign the role around AI, not bolt AI onto the role.
Key Takeaways
The three doomed adaptation strategies
- Master AI tools more than your peers. Bandaid: AI proficiency is a commodity within months.
- Deepen your domain expertise. Bandaid: the AI's expertise rises with the model; your relative differentiation declines.
- Lean on soft skills. Bandaid: soft skills assume a role exists for them to operate in.
All three are defensive reactions inside the existing role. The role itself is the thing being eaten.
The real adaptation: redesign the role
- Identify constraints that previously bounded what you could do — too few campaign variants, too slow analysis, single-threaded review. Now they are no longer constraints.
- Use agents to remove those constraints; reposition yourself as the orchestrator and judgment layer at the new scale.
Two illustrative cases
- Jane, research analyst: $250/hour role compressed to $150/hour as an agent did 80% of her output in 90 minutes. The hours billed didn't drop — the hourly rate did.
- Marcus, marketing strategist: pivoted from execution to running 50 simultaneous campaign variants — work formerly requiring 12 people. He climbed the value stack rather than holding ground inside it.
The systemic incentive problem
- Companies harvest the productivity gain immediately. They have weak incentives to retrain workers for roles that do not yet exist.
- Tegze: "The meta-skill is learning to continuously spot the next evolution and position yourself at the edge of what's newly possible." This pushes the redesign burden onto the worker.
Mental Models
- Second-Order Thinking — Jane is doing the same job; the second-order effect is a quiet hourly-rate compression
- Incentives Matter — the firm captures productivity; the worker bears the redesign cost
- Inversion — instead of "how do I keep doing my job?" ask "what was previously impossible that I could now do?"
- Jevons Paradox — Marcus running 50 campaigns is the labor-market Jevons: efficiency increases the quantity of work that the few empowered workers absorb
See also
- Shipping at Inference-Speed (Steinberger) — the engineering version of Tegze's redesign argument
- The Beauty of Slow Research Assistants (Fletcher) — Fletcher and Tegze are the same observation in different domains: the bottleneck has moved to you
- Something Big Is Happening (Shumer) — Shumer on what serious AI use looks like; Tegze on what failing to do it looks like
- Tyler Cowen on AI and Academia — same dynamic in academia; tenure simply postpones the shrinkage
- AI-Native Software Development — the engineering specialization of Tegze's general thesis
Source
Jan Tegze, Recruitment Economy Substack, 2026 Original article