The AI-Augmented Services Economy

For decades, the services economy scaled through labour arbitrage. Human time converted directly into revenue. Accounting, legal drafting, research, reporting, compliance, consulting deliverables, and management coordination followed a simple rule: hire more people, generate more output, collect more fees.

Generative AI is compressing this model through two mechanisms.

The macro effects are material. McKinsey estimates that generative AI alone could add 0.1–0.6 percentage points annually to global labour productivity growth through 2040. Combined with other automation technologies, the contribution rises to 0.5–3.4 percentage points annually. [1]

Yet adoption remains shallow. McKinsey's 2024 survey indicates that more than two-thirds of organisations globally report using AI in at least one business function, but only a small minority describe their use as mature or scaled. [2] High adoption combined with low maturity points to a significant reservoir of unrealised operating leverage — and a widening gap between firms that redesign workflows and those that layer tools over unchanged processes.

This gap will not close evenly.

To understand what AI disrupts, it helps to reframe what services actually are.

Professional and commercial services are not primarily defined by the expertise they contain. They are labour architectures: organisational structures that convert human time, judgment, and coordination into billable output.

Services account for roughly two-thirds of global GDP and represent the majority of value added in advanced economies. [6] Any sustained shift in the productivity of service delivery is therefore macro-relevant — not a sectoral adjustment, but a structural repricing of how a dominant share of economic output is produced.

At firm level, BIS research on European companies finds that AI adoption is associated with approximately 4% higher labour productivity on average, driven largely by capital deepening, without significant short-run negative employment effects. [7] This is consistent with a transition already underway: from labour expansion as the primary growth lever to tool-enabled leverage as the basis for margin and throughput.

The pattern is not sudden collapse. It is gradual repricing — from labour arbitrage to leverage arbitrage. For allocators, the question is not whether services change. It is which service categories sit on which side of that transition.

AI compression does not operate in isolation. It coincides with a generational shift in how services are purchased and valued.

Younger cohorts are subscription- and interface-native. They expect transparent pricing, productised experiences, and digital-first delivery — particularly in information-rich services where traditional models rely on opaque fee structures and relationship-driven engagement.

This behavioural shift is visible in financial advisory. Robo-advisory revenues were estimated at approximately USD 6.6 billion in 2023 and are projected to grow rapidly through 2030. [4]

Meanwhile, analysis by the City of London Corporation and KPMG projects that AI could contribute up to £35 billion in value and materially increase productivity within UK financial and professional services by 2030. [5]

The implication is not that advisers disappear. It is that fee compression accelerates in segments where cognitive tasks are standardisable and value is not strongly linked to regulated accountability or bespoke complexity.

As cognition becomes cheaper, scarcity migrates.

In an economy where AI can draft, analyse, summarise, and coordinate at near-zero marginal cost, binding constraints shift to domains where software alone cannot deliver:

• Regulated sign-off and liability
• Physical presence and site access
• Safety-critical execution
• Trusted local relationships
• Coordination under real-world constraints

In industrial and execution-heavy services, the deliverable is not the report. It is the accountable act of doing — under conditions where regulation, safety, and physical presence define the service boundary.

AI improves planning, documentation, compliance workflows, and communication within these services. It does not eliminate the requirement for accountable human execution where regulation and risk demand it.

The most compelling near-term application of AI in services is not substitution. It is margin expansion in businesses where physical execution is the core deliverable and back-office drag is the binding constraint on throughput.

In these businesses, AI-enabled automation reduces:

• Scheduling friction
• Administrative overhead
• Reporting time
• Compliance paperwork
• Internal coordination costs

When back-office drag falls, throughput per operator increases and overhead per unit of revenue decreases — without proportional headcount growth.

Grand View Research estimates that the global field service management software market could reach approximately USD 11.8 billion by 2030, reflecting sustained demand for digitised coordination layers. [12]

In fragmented verticals with recurring demand and regulatory constraints, AI-enabled coordination creates a structural advantage for consolidation. Platforms that combine standardised workflows, embedded automation, and proprietary operational data can scale without the linear cost curve that historically constrained service businesses.

The value resides not in proprietary AI models — which will be widely accessible — but in the integration of execution platforms with disciplined AI-enabled workflows and compounding data capture. This is leverage applied to scarcity.

The trajectory of AI capability reinforces this thesis.

Agentic AI — systems capable of autonomously executing multi-step workflows such as scheduling, email handling, and process coordination — is maturing rapidly.

In early 2026, the open-source agent project OpenClaw drew significant attention when its creator, Peter Steinberger, joined OpenAI. Reuters reported that OpenClaw would transition into an independent foundation supported by OpenAI. [8][9] At the same time, several technology firms restricted internal use of similar tools due to security and governance concerns. [10]

Adoption remains uneven.

McKinsey research documents high experimentation but limited scaled maturity across industries. [2][3] OECD research indicates meaningful but incomplete SME adoption, with capability and skills gaps as primary constraints. [11] BIS evidence suggests that productivity gains concentrate in firms that pair AI adoption with genuine organisational change. [7]

The competitive gap between operators who redesign workflows around AI and those who layer it superficially will widen.

For allocators, the distinguishing variable is not whether a portfolio company uses AI. It is whether AI is embedded in the operating model or applied as a surface-level addition.