Agent Economics: Consulting’s Next AI Battleground

Summary: EY’s new Head of Agent Economics signals that consulting’s AI competition is shifting from adoption towards measurable commercial value. Firms must connect agent costs with delivery, pricing, talent and client outcomes while preserving the human judgement on which consulting depends.

Consulting’s AI race is moving from capability to commercial discipline.

EY’s decision to hire a Head of “Agent Economics” shows where that race is heading. The role will oversee its agentic workforce and lead a new AI value realisation office.

Its purpose is not simply to reduce technology spending. It is to show EY’s leadership where AI is being used, what it costs and whether it creates material business value.

The title may sound unusual, but the problem will soon be familiar across professional services. Firms have spent the first phase of generative AI competing on access, experimentation and adoption.

They have launched platforms, trained employees and embedded assistants into everyday work. The next phase is about proving that this activity improves margins, strengthens delivery and creates better client outcomes.

More AI does not automatically create more leverage.

A Second Leverage Model

The traditional consulting pyramid is built around human leverage. Senior people sell work, frame the problem and manage the client relationship.

Larger junior teams conduct research, analyse data and prepare recommendations. Their work provides delivery capacity and the apprenticeship through which future leaders learn the profession.

AI agents introduce a second form of leverage. They can perform parts of that execution layer at speed and allow smaller teams to produce more.

The commercial potential is significant. Firms could shorten projects, reduce low-value production and concentrate senior time on decisions, relationships and implementation.

But AI leverage behaves differently from human leverage. People have salaries, utilisation targets and relatively visible capacity, whereas AI consumption varies with usage, model choice and task complexity.

An agent may save hours while consuming expensive tokens. A frontier model may produce a better answer without improving the outcome for the client.

Faster output can also create more review, governance and quality-control work. Unless the whole workflow changes, productivity gains may simply move effort elsewhere.

Agent Economics and the Cost of Consumption

Conventional consulting measures do not make these economics easy to see. High adoption can look like progress even when AI costs increase faster than the value created.

Time saved can sound impressive without improving project margin or shortening delivery. Usage figures reveal activity, but they do not prove value realisation.

EY has introduced token budgets in areas of heavy use. It is also using model routers to direct people towards the appropriate level of capability and token cost.

McKinsey is confronting the same issue through “circuit breakers” that flag heavy token use. Its CFO captured the principle neatly: you do not need a Ferrari for the school run.

You do not need a Ferrari for the school run.McKinsey’s CFO, on matching the model to the task

These controls are not an argument for restricting AI. They show that model choice, agent design and computing consumption are becoming management decisions.

The question is no longer whether employees are using AI. Leaders need to know whether they are using the right systems for work that justifies the cost.

Ownership Changes the Organisation

Agent economics exposes an organisational gap because AI agents do not sit comfortably within any existing function. Technology can manage infrastructure, while finance can track expenditure and returns.

HR can govern people and workforce planning, risk teams can set controls, and business leaders can redesign workflows. None can manage the whole system independently.

An agentic workforce cuts across headcount, skills, accountability, intellectual property, client risk and commercial performance. Decisions in one area quickly create consequences elsewhere.

A technically efficient decision may weaken development pathways for junior consultants. Removing routine work could remove the apprenticeship through which future Partners learn to structure problems and exercise judgement.

An agent may improve internal productivity yet create little value if the firm continues to sell and staff projects in the same way. The technology changes, but the economics remain untouched.

That is why EY’s value realisation office matters more than its unusual title. It creates accountability for connecting AI investment to operating outcomes.

The function must connect workflow, commercial value and organisational change.

Agent Economics as a Leadership Discipline

The Head of Agent Economics should not become an AI “beancounter-in-chief” focused on token consumption. The role creates value only if it addresses more consequential questions.

Which work should agents perform, and where should they augment consultants? Which models are sufficient for different levels of complexity and risk?

Leaders must decide how teams should be redesigned, how savings should be shared with clients and what new services become possible. They must also identify which human capabilities become more valuable as execution is automated.

These are business-model questions, not technical ones. They require leaders who can work across functions without allowing the AI agenda to become trapped inside one.

Agent economics will fail if finance sees AI only as a cost. It will also fail if technology treats adoption as the objective or business leaders protect existing delivery structures.

The role needs authority, cross-functional credibility and access to senior leadership. It also needs enough understanding of consulting delivery to distinguish genuine value from attractive internal metrics.

This changes the hiring brief. Strong candidates may combine AI operations, workforce design, financial discipline and experience of delivering client value.

Client Value Becomes the Test

The most important implication is external. Consulting firms are advising clients to adopt agents, redesign work and capture productivity gains.

They cannot credibly make those promises without understanding agent economics inside their own businesses. Their internal transformation will become evidence for, or against, their client proposition.

Clients will ask how much work has been removed and why an AI-enabled project needs the same team. They will question whether faster delivery should reduce the fee and who remains accountable for agent-generated work.

They will also ask what happens to quality when lower-cost models are used. Above all, they will expect measurable improvements rather than a description of the technology deployed.

These questions will challenge the billable-hour model, but they do not make consulting less valuable. They shift value away from production volume and towards decisions and implementation.

When analysis becomes faster, the premium moves towards problem framing, judgement and specialist expertise. Stakeholder confidence, organisational navigation and accountability become more important, not less.

The winning firms will remove low-value production while concentrating senior attention where it matters. They will build smaller expert teams, deploy technology selectively and price more confidently around outcomes.

Talent Must Change Too

Firms must preserve the learning and challenge that produce good consultants. Automating routine work without rebuilding development would improve today’s efficiency at the expense of tomorrow’s leadership pipeline.

Firms may need fewer people performing repeatable analysis but more who can frame ambiguous problems and challenge machine output. They will need consultants who can translate faster insight into organisational action.

Partners will need to understand delivery economics as well as client relationships. Technical leaders will need commercial judgement.

EY may be early in naming the role, but it is not alone. Accenture has launched its own token-spend practice, and once agents enter everyday delivery, their economics cannot remain hidden inside technology budgets.

Deploying more agents will not create an advantage by itself. The differentiator will be better delivery, stronger margins and measurable client value.

Why Agent Economics Changes the Race

Consulting’s AI race is moving beyond adoption and scale into a more demanding phase of commercial discipline. The winners will not be the firms deploying the most agents or consuming the most powerful models. They will be those that understand where AI improves delivery, strengthens margins and creates measurable client value.

Just as importantly, they will know where agents introduce hidden costs, weaken development or require the judgement and accountability that only experienced consultants can provide.

Ben Appleton

Ben Appleton is the founder of Strat-Bridge, a specialist executive search partner to the management consulting industry. He works with global consulting firms and senior leaders across the UK, Germany, Switzerland, and beyond — helping them build capability at the Partner and Director level.

Connect with Ben on LinkedIn or email ben@strat-bridge.com.