Research framework · conditions to Q1 2026

Sectoral deployment dynamics

GenAI's labour-market effects are not uniform across industries. This framework maps ten sectors on three dimensions: how far deployment has progressed, how elastic demand is for what the sector produces, and how mature the implementation has become. Demand elasticity is the variable that decides whether productivity gains show up as expansion or as displacement.

Demand elasticity

Type A: elastic (expansion absorption)

Type B: inelastic (displacement)

Type C: mixed (restructuring)

Bubble size

AI-native firm formation intensity

Dimension III: implementation maturity curve

Phase I

Naïve substitution

Direct human-for-AI replacement preserving existing workflows. Fails on edge cases, tacit knowledge, and contextual judgment.

Klarna 2024: AI chatbot replaces agents, quality collapses on complex cases, partial reversal follows.

Phase II

Workflow redesign

Hybrid architectures: AI handles volume, humans handle exceptions. Workflows restructured around AI comparative advantages.

Klarna 2025 hybrid model; enterprise Copilot deployments with human review gates; Salesforce Agentforce triage.

Phase III

Agentic integration

Multi-step autonomous agents decompose complex tasks, self-verify, and escalate only at low confidence. Eliminates the human coordination layer.

Sierra AI autonomous CX agents; Harvey agentic legal workflows; Cursor and Devin agentic coding.

Phase IV

Autonomous operation

End-to-end AI workflows with human oversight scaling sublinearly with output volume. Periodic quality auditing replaces continuous supervision.

Emerging: fully autonomous code-generation pipelines; AI-native legal document factories; lights-out customer resolution.

Dimension II: demand elasticity and employment equilibria

Type A: elastic demand

Expansion absorption

Cost reductions unlock large latent demand: the access-to-justice gap in legal services, physician shortages in healthcare, tutoring access in education. Market expansion absorbs the productivity gains.

Employment outcome: stable or growing total employment, with a strong compositional shift. Fewer junior execution roles, more judgment-intensive and client-facing ones. The career pipeline restructures.

Legal services Healthcare Education

Type B: inelastic demand

Displacement equilibrium

Output is a derived cost centre; nobody wants more support tickets or data entry. Demand is roughly fixed regardless of production cost, so productivity gains flow to the bottom line.

Employment outcome: permanent structural displacement. Headcount stabilises at a fraction of pre-AI levels, and residual human roles are limited to complex exceptions and oversight.

Customer support Data entry Translation Back-office admin

Type C: mixed elasticity

Compositional restructuring

Elastic demand for the high-value insight layer (strategic advisory, complex analysis) paired with inelastic demand for the routine production layer (report generation, compliance, data reconciliation).

Employment outcome: net employment roughly stable, with a strong compositional shift. The routine layer contracts while the high-value layer expands. Junior execution roles are hollowed out; senior judgment roles persist.

Financial services Consulting Software development Creative industries

Figure: sectoral deployment dynamics model. Each sector is positioned by its current deployment stage (x-axis) and by the employment-outcome trajectory that its demand elasticity implies (y-axis). Bubble size reflects AI-native firm formation intensity, and the trailing arrow indicates the pace of movement. The framework integrates Perez's installation and deployment periodisation with Arthur's combinatorial buildout algorithm, adding demand elasticity as a meso-level variable that governs differentiated sectoral employment equilibria. Dashed vertical lines mark the four deployment stages; horizontal bands indicate the demand-elasticity types. Data reflect conditions as of Q1 2026.