Skip to content
AI Leadership

Matt Shumer: 'your job as a knowledge worker will no longer be...

In February 2026, Matt Shumer (OthersideAI, a hands-on AI reference) published an open letter challenging: knowledge workers will be replaced in scope within less than...

By Paulo Castello6 min read

Matt Shumer: 'your job as a knowledge worker will no longer be executed by you in less than 5 years'

In February 2026, Matt Shumer — founder of OthersideAI and a hands-on reference in AI — published an open letter about the future of the knowledge worker. An uncomfortable message, but mathematically hard to refute for anyone using AI hard.

"This guy Matt Shumer has been a reference over the past few years, closely tracking every step of AI's evolution. I've watched dozens of his videos reviewing new models, articles, and techniques. The guy is super hands-on. His final message: 'Your job (as a knowledge worker) will no longer be executed (in its entirety) by you in less than 5 years'."

— Paulo Castello, February 2026

Who is Matt Shumer and why Brazilian C-level needs to know

Matt Shumer is the founder of OthersideAI. Not a "thought leader" who gives talks — he is a practitioner. The type who tests every release of GPT, Claude, Gemini on the day it drops. Who does a teardown of every new framework. Who publishes technical reviews with real hands-on work.

In Portuguese, he is not widely known. In English, he is an absolute reference for anyone tracking AI evolution with depth.

When Matt Shumer publishes something, it is worth stopping to read. Not for institutional authority (he is not Karpathy, he is not Altman), but for the technical credibility accumulated in thousands of publicly documented hands-on hours.

In February 2026, he published an open letter that went viral globally. The phrase that landed:

"Your job (as a knowledge worker) will no longer be executed (in its entirety) by you in less than 5 years."

What he really means (and what he does not)

Naive reading: "I'll lose my job in 5 years."

That is not it. Correct decomposition:

  • "no longer be executed (in its entirety) by you" = the execution component of your work migrates to AI
  • "in less than 5 years" = a short window, not decades
  • implicit implication: you remain relevant if you migrate toward the direction component; you become obsolete if you stay attached to the execution component

Practical translation:

  • A financial analyst stays relevant by migrating to direction (defining the analysis hypothesis, judging conclusions, deciding on investments) — becomes obsolete if staying attached to building the Excel model
  • A programmer stays relevant by migrating to architecture, strategic technical decisions, critical code review — becomes obsolete if staying attached to writing boilerplate
  • A lawyer stays relevant by migrating to legal strategy, negotiation, judgment in novel cases — becomes obsolete if staying attached to drafting standard contracts
  • A designer stays relevant by migrating to product vision, art direction — becomes obsolete if staying attached to producing catalog illustrations

Pattern: the executing component of every knowledge function migrates to AI. The directing component remains human.

Why Brazilian C-level needs to absorb this

It is not just about the professionals on your team. It is about you.

Sam Altman has already stated explicitly that "not even CEOs will be spared."

Components of the CEO role that will migrate to AI in the next 5 years:

  • Financial report analysis
  • Board presentation preparation
  • Executive communication drafting
  • Market research and benchmarking
  • Operational risk analysis

Components that remain human:

  • Judgment under radical ambiguity (M&A decisions, strategy shifts, existential crises)
  • Deep relationships (investors, board, strategic partners, core team)
  • Moral decision-making in edge cases
  • Long-term direction-setting
  • Leadership through turbulence

The CEO who continues to identify as "the one who put together that spectacular presentation" will become obsolete. The CEO who identifies as "the one who decided to make the AI First transition, led the team through the process, and captured operational leverage" remains relevant for decades.

The practice at Fhinck that confirms the thesis

Paulo Castello comments on the post:

"We are living this in practice at Fhinck and I agree entirely."

Fhinck went through — in 24 months (2023–2025) — exactly the scenario Shumer describes in 5 years.

50 people → 6 people. Not through a cruel mass layoff, but through a simple observation: the executing component of many functions migrated to agents. The 6 who remained are precisely those who migrated toward the directing component.

And he notes:

"For the first time in my professional life, I struggled (and felt insecure about) advising a relative about the future of their career (because I can see that what they know about career and work will be fully taken over by AI in the coming years)."

If the CEO of Fhinck — who lived the transition — struggles to advise a relative, the average Brazilian professional is lost. This is why the topic needs to become an agenda item for boards, management teams, and dinner tables.

The quick self-test (do it today)

List your 10 main weekly activities. For each one, classify:

  • E — predominantly EXECUTION (produce, draft, analyze, process)
  • D — predominantly DIRECTION (decide, judge, connect, lead, define)

Count:

  • 7+ E: high risk in 36–60 months. Urgent migration needed.
  • 4–6 E: medium risk. Planned migration over the next 24 months.
  • 0–3 E: low risk. But keep migrating — do not relax.

Do this exercise honestly. Repeat every 6 months to measure migration.

The 12-month plan to survive the Shumer shift

  1. Today: do the self-test above. Note the number of E vs D.
  2. Month 1–3: identify 1 E activity that can be delegated to your own agent (built by you). Implement it.
  3. Month 4–6: identify the D skill that is growing most in value in your sector. Invest 5h/week in learning it.
  4. Month 7–9: redesign your weekly routine. Reduce time in E, increase time in D.
  5. Month 10–12: redo the self-test. If the E:D ratio improved, you are on track. If it got worse, recalibrate.

Conclusion

Matt Shumer's open letter is uncomfortable because it is precise. In 2026, those deep in hard AI recognize the thesis. Those who are not think it is an exaggeration.

Time will tell who read it right. But the cost of being wrong through optimism ("Shumer is exaggerating") is greater than the cost of being wrong through prudence ("I'll prepare even if he's exaggerating").

Fhinck has built an institutional culture aligned with this thesis — weekly Sharpening the Axe, agents replacing execution within defined scopes, humans focused on direction. If you want to think about how to apply this in your team, schedule a conversation.


Keep reading

Matt ShumerOthersideAIknowledge workerfuture of workopen lettercareer

Paulo Castello

CEO & Founder, Fhinck

Led the transition of Fhinck from a traditional Task Mining company to AI First — from 50 to 6 people with double the revenue.

View profile →

Want to see this in your operation?

Schedule a demo and discover the 600% ROI in under 12 months.

Schedule a demo