Will AI Agents Replace Employees? The Truth Nobody Tells You

In 2023, Klarna — the Swedish fintech giant — announced it had slashed its headcount from 3,800 to 2,000 employees by replacing human support with AI agents. Two years later, CEO Sebastian Siemiatkowski publicly admitted: service quality dropped, and the company started hiring humans again.

That reversal wasn't a failure admission — it was a lesson learned. The truth about AI and jobs is more nuanced than the extremes dominating public discourse: neither "AI will eliminate all jobs" nor "AI won't change anything." The real answer lies in the data, revealing a scenario of selective displacement with net job creation — but with transition costs unequally distributed.

What the Data Actually Says (Not the Hype)

In November 2025, the McKinsey Global Institute published a comprehensive study on automation in the US labor market. The numbers are clear:

57% of hours worked in the US can be automated with technology available today
40% of jobs are in highly automatable functions
94% of employees and 99% of C-level executives already use some form of AI at work

The limiting factor isn't technological capacity — it's adoption and deployment velocity. Companies slow to implement AI agents fall behind.

The World Economic Forum's Future of Jobs Report 2025 projected the global scenario through 2030:

Metric	Value
Jobs displaced	92 million
New jobs created	170 million
Net change	+78 million jobs

Goldman Sachs adds: ~300 million jobs globally will be affected by AI — but affected ≠ eliminated. Approximately 6-7% of the American workforce (11 million people) will be displaced, yet 85% of US job growth since 1940 came from creating new tech functions.

For emerging markets like Brazil, the International Labour Organization estimates that 2.38% of jobs are at risk of total automation — lower than developed nations due to lower digital exposure.

Real Cases: Companies That Replaced and What Happened

Klarna — Replacement → Retreat

Klarna became the most cited case of aggressive automation. Hiring freeze in 2022-2023, AI chatbot handling work equivalent to 800 employees. Initial result: significant cost reduction.

Plot twist: In 2025, customers started complaining about service quality. Satisfaction dropped. The company resumed hiring humans for support — with remote/contract models.

Lesson: Total automation of customer service works for volume, not satisfaction. AI agents excel at FAQs, but complex issues require human empathy.

IBM — AskHR with Caveats

IBM laid off ~8,000 employees and replaced much of HR with the "AskHR" agent. The system manages 11.5 million annual interactions with less than 5% human supervision, resolving 78% of queries without escalation.

Caveat: AI failed at tasks requiring empathy and subjectivity — they rehired part of the team.

CEO Arvind Krishna: "Our total employment increased because AI frees investment for other areas."

Lesson: AI agents automate processes; humans handle complex cases.

Dukaan — Total Success

Indian e-commerce platform Dukaan replaced 27 support agents with a ChatGPT-based system. 99% cost reduction in support with 85% satisfaction maintained.

Context: SME with predictable demand and repetitive questions. Works well here.

Lesson: For small and medium businesses with standardized query volume, AI agents are margin-transformative.

GitHub Copilot — Amplification Without Layoffs

Microsoft and GitHub widely adopted Copilot for internal and customer developers. Result: developers 55% faster at coding tasks — without mass layoffs.

Lesson: AI as a tool for individual capacity amplification. "1 dev equals 10" with the right tool.

Corporate Regret

An Orgvue survey of business leaders revealed something concerning:

4 in 10 executives laid off employees to implement AI
55% regretted it afterward
25% didn't know which functions would benefit most from AI
30% didn't know which workers were most at risk

Conclusion: Replacing without planning generates regret. The right strategy is mapping automatable tasks before cutting people.

What AI Agents Are Automating Now

Functions with high risk of replacement by AI agents:

Function	Risk Level
Manual data entry	95%
Routine customer service	80%
Administrative assistants	75%
Legal/medical secretaries	63-75%
Simple contract review	70%+
Standardized report generation	80%+
Level 1 technical support	75%+
Basic content moderation	85%+

SSRN estimates that 7.5 million data entry and administrative jobs could be eliminated by 2027.

Common characteristics: Repetitive tasks, information-based, low ambiguity, no empathy or moral judgment required.

What AI Agents CANNOT Replace

Functions with low automation risk:

Clinical healthcare: Doctors, nurses, caregivers (human touch + critical decisions)
In-person education: Teachers (relationships, motivation, socioemotional adaptation)
Strategic leadership: CEOs, directors (responsibility, vision, culture)
Complex B2B sales: Trust relationships, high-value negotiation
Law (litigation): Oral argumentation, equity judgment
Original art and creativity: Conceptual design, unique authorial narrative
Complex physical work: Construction, installations, maintenance
Psychology and therapy: Deep empathy, therapeutic bond

Why they resist: Empathy, decision-making with moral ambiguity, original creativity, physical presence, legal responsibility, relational trust. AI agents process information — they can't judge equity, negotiate human conflicts, or create art with authorial intent.

The New Professions AI Is Creating

The WEF projects 170 million new functions by 2030. Some already exist:

Profession	Trend
Prompt Engineer	Already hiring ($100-300k/year in the US)
AI Trainer / Data Labeler	Thousands of global openings
AI Agent Orchestrator / AgentOps	Emerging, high demand 2025-26
AI Ethics Officer / Governance	Growing regulation (EU AI Act)
AI Product Manager	Top-growing title 2025
Knowledge Architect	Structuring bases for RAG/LLM
Conversation Designer	Designing AI agent dialogue flows
Human-AI Collaboration Lead	New role in Fortune 500
AI QA Engineer	Testing model behavior

Goldman Sachs records +216,000 data center construction jobs since 2022 — physical infrastructure to sustain the digital layer.

Message for professionals: The question isn't whether AI will change your job. It's whether you'll prepare for the functions being born.

What This Means for SMEs

For small and medium enterprises, the picture differs from large corporations:

Lower digital exposure: Fewer processes eligible for automation
Smaller AI budget: ROI must be clear from day 1
Greater need for personalized service: SME customers want to talk to humans

INOVAWAY recommendation: Don't think "replacing people." Think automating repetitive processes to free your team for higher-value activities.

Practical framework:

Identify tasks taking 20%+ of your team's time
Assess if they're standardized and low-ambiguity
Implement AI agents for these specific tasks
Reallocate humans to complex handling, relationships, and strategic decisions

The focus isn't cutting — it's increasing productivity without proportionally increasing headcount.

Conclusion

The answer to "Will AI agents replace employees?" is: They replace tasks, not entire roles.

The data shows:

92 million functions will be displaced by 2030
170 million new functions will be created
Net change: +78 million jobs
But the transition is painful for those unprepared

Companies that cut without planning regret it. Companies that strategically automate processes gain competitive advantage.

** The real question isn't whether AI will change your work. It's whether you'll be ready to lead the change — or be blindsided by it.**

Next Steps

Want to discover which processes in your company can be automated with AI agents without compromising service quality? Talk to INOVAWAY — we analyze your workflow and identify real automation opportunities.

Related posts:

INOVAWAY Intelligence