The Automation Curve in Commerce Points to a Bigger Opportunity in Services

Based on McKinsey's article "The automation curve in agentic commerce."

The future of AI commerce is unlikely to arrive all at once.

It will probably unfold in stages.

That is one of the most useful ideas in McKinsey's framing of the "automation curve" in agentic commerce. Instead of assuming a sudden jump from human-driven shopping to full autonomy, the better model is gradual delegation. Users will hand over some parts of the journey before others. Trust, risk, complexity, and context will shape how far that delegation goes.

That is important because it gives a more realistic view of how markets will change.

Delegation grows in layers

Not every buying decision is equally easy to automate.

Some decisions are low risk, repetitive, and easy to define. Others are emotional, high stakes, ambiguous, or dependent on nuanced judgment. That means the role of AI agents will expand unevenly.

In some cases, AI may only help users narrow the field. In others, it may assemble options, compare tradeoffs, and suggest action. In still others, it may be authorized to execute within certain boundaries.

That progression matters because it suggests where infrastructure has to improve first.

The winners will be the businesses and platforms that can support multiple levels of delegation, not just one idealized end state.

Machine readability becomes a hard requirement

As delegation rises, businesses need to become easier for agents to understand and act through.

If catalogs, constraints, policies, service levels, and transaction paths are not machine-readable, AI systems will struggle to include them in decision flows. In an agent-mediated market, invisibility can come not from bad branding but from bad structure.

This is especially important because AI agents operate on the same internet as humans. They browse websites, interact with APIs, evaluate policies, and compare options across many sources. That means businesses do not get a separate transition period. The existing web becomes the training ground for delegated commerce.

Why the opportunity is even bigger in services

This logic becomes even more important in services.

In products, delegation can often move across a relatively structured path: discovery, comparison, basket building, payment. Services are harder because the transaction itself is often not fully known at the start.

Can the provider do the job? When are they available? What is the real scope? How urgent is the issue? What is the right price? What happens if the first provider declines?

These are not edge cases in services. They are the normal case.

That means service markets are one of the places where the automation curve becomes more operationally demanding. AI agents cannot simply rank and recommend. They need an infrastructure layer that can help determine feasibility and move toward commitment.

What this means for Aune

This is where Aune becomes especially relevant.

Aune is built for a world in which users and AI agents delegate more of the service journey, but not necessarily all of it at once. The goal is not to assume perfect autonomy from day one. The goal is to support the real path from intent to execution under real constraints.

That means helping structure service supply in a way that supports multiple levels of delegation: finding providers, comparing options, checking fit, negotiating terms, securing commitment, and managing fallback paths when needed.

This is exactly the kind of execution layer that services will require as agent-mediated commerce matures.

If product markets need better machine readability, service markets need that plus transaction logic.

That is the opportunity Aune is built around.

Final thought

The automation curve is a useful reminder that the future will not arrive in one leap.

Users will delegate gradually. Trust will build unevenly. Different categories will move at different speeds.

But the direction is clear.

As AI agents mediate more of commerce, the categories that require real-world coordination will need a deeper layer of infrastructure.

In services, that layer is not optional.

It is the path from visibility to execution.