Summary
The narrative that AI is wiping out the workforce is a dangerous distraction. Recent data from Lightcast and Gartner reveals that less than 1% of recent layoffs are genuinely caused by AI displacement; the rest is a correction of COVID-era overhiring. The economic reality is shifting from cost-per-employee to cost-per-task, where agents can handle customer service interactions for pennies on the dollar compared to humans. This forces a structural change: team compression. Fewer people doing more work, but the work is reorganized around what humans do best. The real opportunity lies in the net-new roles emerging to manage these systems, like AI trainers, and the workflows that sit in the gap between what’s too simple to need a human and what’s too complex to fully automate.
Research
The ‘AI Layoff Illusion’ paper from Lightcast and Gartner provides the data to kill the doom narrative. It shows that less than 1% of recent cuts are genuine AI displacement; the rest is a correction of COVID-era overhiring. The economic reality is shifting from cost-per-employee to cost-per-task, where AI agents—specifically in customer service—can drop interaction costs to $0.25 versus $3-$6 for humans. This forces a reorganization of team structures: compression. Fewer people doing more work, but the work itself is reorganized around what humans do best. The narrative of total replacement is a distraction. The real opportunity lies in the net-new roles emerging to manage these systems, like AI trainers and evaluators, and the workflows that sit in the gap between what’s too simple to need a human and what’s too complex to fully automate.
Books
From Deerfield Green’s library of long-form research — books written to give practitioners the economic models, case studies, and strategic depth that whitepapers and blog posts can’t. Here’s what’s relevant this week.
AI Augmentation Creates Net-New Roles
The workforce displacement narrative focuses on roles that AI eliminates or diminishes. Less discussed — but economically significant — is the emergence of net-new roles that exist only because AI makes them possible. AI trainers and evaluators are humans who assess AI outputs, provide feedback, and curate training data. This work did not exist before large language models, and it now employs hundreds of thousands of people globally — mostly in lower-wage sectors. This highlights a crucial economic reality: AI creates a new class of infrastructure workers who maintain the models that drive the economy.
Source: books/enterprise-ai-economics/chapters/ch06-organizing-ai-workforce.md
The Before and After: How AI Reshapes Team Structures
Most leaders talk about AI and workforce in the same breath as ‘augmentation’ — this comforting notion that AI will make everyone more productive and nobody will lose their job. That framing is politically convenient and operationally dishonest. What AI actually does to team structures is compression: fewer people doing more work, with the work itself fundamentally reorganized around what humans do best. This isn’t about making a typist faster; it’s about eliminating the typist entirely and redefining the role to require high-level strategic oversight. The workforce is shrinking in volume but expanding in cognitive demand.
Source: books/before-you-buy-the-robot/chapters/ch03-worker-skillsets.md
Articles
Curated from recent reporting and analysis across the industry. These are the pieces we think cut through the noise.
Human-on-the-Loop: Enterprise Workflow Control
The two models, human-in-the-loop (HITL) and human-on-the-loop (HOTL), determine how much control you keep over each decision and how fast the workflow moves. HITL requires human approval at every step, creating a bottleneck that limits speed but ensures accuracy. HOTL allows the system to operate autonomously within a set of guardrails, stepping in only when confidence drops below a threshold. For enterprise leaders, the choice isn’t between automation and humans; it’s between total control and speed. The most effective strategies often sit in the middle, using HOTL to handle the high-volume, low-stakes decisions while reserving HITL for critical compliance and liability issues.
Source: https://www.elementum.ai/blog/human-in-the-loop-vs-human-on-the-loop
The Economics of AI Agents: Cost Per Task vs Cost Per Employee
AI agents are introducing a radically different economic model based on cost per task, where businesses pay only for specific outputs rather than ongoing employment. This shifts the ROI calculation from a headcount budget to a utility bill. When you compare the cost of an AI agent handling a customer service interaction — often $0.25 or less — against a human agent requiring $3-$6 in operational expenses, the math is undeniable. The break-even point typically occurs after 40,000 to 60,000 interactions annually. This forces companies to move from ‘how many people do we need?’ to ‘what tasks do we need to automate?’
Source: https://www.businessplusai.com/blog/the-economics-of-ai-agents-cost-per-task-vs-cost-per-employee
White Papers
Deerfield Green publishes original research on the forces reshaping labor markets, token economics, and enterprise adoption curves. These excerpts are drawn from that ongoing work.
The AI Layoff Illusion: How COVID Overhiring Became ‘AI Efficiency’
The target audience for this whitepaper is enterprise leaders evaluating whether AI-attributed workforce reductions reflect genuine automation displacement or narrative reframing. The key takeaway is that less than 1% of recent layoffs are genuinely caused by AI. The paper argues that ‘AI efficiency’ is serving as narrative cover for restructuring that was inevitable regardless of AI advances. Companies are using the AI narrative to justify cuts they needed to make to fund massive AI capital expenditures, creating a false correlation between automation and job loss.
Source: whitepapers/ai-layoffs-vs-covid-overhiring/outline.md
AI Layoffs vs COVID Overhiring: Research Document
The thesis of this research is that tech layoffs attributed to AI are largely a correction of COVID-era overhiring (2020-2022). Genuine AI displacement is beginning at the margins — particularly in customer service and content translation — but accounts for a small fraction of total cuts. When Amazon cites AI while cutting 14,000 corporate roles, the real driver is often cost optimization to fund $650B in AI capex. The ~55,000 AI-cited figure for 2025 should be treated as a ceiling, not a floor, for genuine AI displacement.
Source: whitepapers/ai-layoffs-vs-covid-overhiring/research.md
Gartner AI Layoff Attribution Analysis
Gartner’s finding that less than 1% of recent layoffs are genuinely caused by AI is a critical data point for executives. This statistic suggests that the fear of mass unemployment is largely unfounded in the short term. However, the paper warns that this is a lagging indicator. As the technology matures and inference costs drop further, the displacement rate will accelerate. The current ‘pause’ is a correction, not a halt to the trend. Leaders need to prepare for a second wave of disruption that will be harder to mask as a market correction.
Source: whitepapers/ai-layoffs-vs-covid-overhiring/research.md
Prototypes
We don’t just write about the future — we build it. Deerfield Green’s prototype lab produces interactive tools that let you stress-test ideas against real data. Here’s what applies to this week’s topic.
Industry Disruption Prediction Market
This interactive tool allows you to speculate on which specific industries will be disrupted first by autonomous AI agents. By aggregating predictions from a network of experts, the market generates a probability-weighted forecast. The prototype visualizes the gap between ‘perceived risk’ (industries everyone fears) and ‘actual economic risk’ (industries with high token consumption and low human oversight). Use this to allocate your AI investment budget toward sectors where the market underestimates the disruption potential.
Source: prototypes/augur/README.md
Autonomous Supply Chain Simulation
This simulation demonstrates how autonomous agents could replace human logistics managers in a complex supply chain scenario. The prototype runs a 12-month forecast comparing a human-managed supply chain (with delays, errors, and manual re-planning) against an agent-managed chain (with continuous optimization). Key metrics include ‘inventory holding costs’ and ‘on-time delivery rates.’ The simulation reveals that while agents reduce costs, they introduce a new vulnerability: the ‘black box’ problem where decision-making is opaque, making it difficult for humans to intervene effectively.
Source: prototypes/augur/README.md
Frameworks
From Deerfield Green’s library of strategic frameworks — structured models for measuring AI value, planning workforce transitions, and sizing transformation initiatives. These are the lenses we use internally, published so you can use them too.
The New AI Workforce: Interactive Reference Board
This framework is an interactive reference board mapping 44+ AI-first job roles emerging across enterprises in 2024–2026. It organizes roles by archetype (Strategy, Engineering, Governance, Operations) and tags them by status: ‘Net-new’ (roles that didn’t exist before agentic AI) and ‘Augmented’ (roles that have been radically upgraded). It provides compensation data and organizational patterns, helping leaders visualize where their current headcount fits into the new structure and identifying gaps in their talent acquisition strategy.
Source: frameworks/the-new-ai-workforce/README.md
Workflow Intent Library: Executive Summary
This framework catalogs 80 canonical workflows across 8 business domains, each tagged with implementation tier, effort size, AI capability pattern, and value vector alignment. The goal is to provide an implementation layer for the AI Value Measurement Framework. It categorizes workflows into patterns like ‘Classification + Orchestration’ and ‘Analysis + Generation,’ mapping them to specific roles such as AP Analyst, Controller, or Revenue Accountant. This allows organizations to standardize their AI adoption and measure value delivery across the enterprise.
Source: frameworks/ai-workflow-intent-library/workflow-library-reference.docx
What’s Next
The agent revolution isn’t coming in a single dramatic moment. It’s arriving one automated workflow at a time, in the gap between what’s too simple to need a human and what’s too complex to fully automate.
References
- [1] The AI Layoff Illusion: How COVID Overhiring Became ‘AI Efficiency’, Whitepaper Outline
- [2] AI Layoffs vs COVID Overhiring: Research Document, Whitepaper Research
- [3] AI Agents vs Human Labor: Market Size, Costs, and Adoption Outlook, DataKnobs Blog