The Spectrum of Delegation

Commerce is not switching from human to machine overnight. It is migrating across five levels. Where is your category?

The Migration, Not the Switch

The most common mistake executives make when confronting agentic commerce is to frame it as a binary: either humans buy, or machines buy. This framing is wrong, and it produces bad strategy.

The transition from human-mediated to agent-mediated commerce is not a switch. It is a spectrum. Different categories, different consumer segments, and different transaction types sit at different points on that spectrum today, and they are all moving — at different speeds, in the same direction.

The Spectrum of Delegation, introduced in Accornero (2026), maps five distinct levels of AI involvement in purchasing decisions. Each level represents a qualitatively different relationship between the human consumer and the AI system. Each level requires a different commercial strategy to compete effectively.

Agentic Commerce, by formal definition, begins at Level 2. Everything below Level 2 is AI-assisted human commerce. Everything from Level 2 upward is, to varying degrees, agent-mediated commerce.

Level 0: Human-Only Commerce

What it looks like: The consumer searches, evaluates, compares, and purchases entirely without AI involvement. Traditional retail, unassisted e-commerce, in-store purchasing without digital aids.

Who is still here: High-touch luxury purchases (bespoke tailoring, art collecting), regulated professional services (legal retainers, private wealth management), and consumers who have deliberately opted out of AI assistance.

Commercial strategy: Traditional marketing applies in full. Brand storytelling, emotional positioning, in-store experience, and human relationship management remain the competitive weapons. This is the world marketing was built for.

The trajectory: Shrinking. Even categories that seem permanently human-mediated are seeing AI encroachment at the research stage. A consumer buying a bespoke suit may still visit Savile Row in person, but they increasingly use AI to research tailors, compare prices, and evaluate options before walking through the door.

Level 1: AI-Informed Commerce

What it looks like: The consumer uses AI systems for research and information gathering, but retains full decision authority. The AI recommends, compares, and summarises. The human decides and purchases.

Examples today: Asking ChatGPT "What are the best wireless headphones under £200?" Using Perplexity to research CRM software. Reading AI-generated product comparisons. Using Gemini to summarise product reviews.

Who is here: The majority of AI-using consumers in 2026. Bain & Company reports that 30 to 45 per cent of consumers already use AI for product research. This is the heartland of current AI-commerce interaction.

Commercial strategy: This is GEO territory. Generative Engine Optimisation governs how you compete at Level 1. The goal is citation, recommendation, and presence in AI-generated responses. Content authority, structured data, domain reputation, and factual verifiability determine whether the AI recommends your product to the human who will then decide.

The trajectory: This is the staging area for agentic commerce. Consumers at Level 1 are building trust in AI recommendations. As trust increases, they delegate more. The path from Level 1 to Level 2 is not a leap. It is a series of small surrenders: "Just buy the one you recommended last time."

Level 2: AI-Guided Commerce (Agentic Commerce Begins)

What it looks like: The consumer defines broad parameters and the AI agent narrows the field to a shortlist. The consumer retains veto power and makes the final selection, but the agent has already eliminated most options. The agent searches, filters, and pre-selects. The human confirms.

Examples today: Amazon's shopping assistant presenting "top 3 options based on your preferences." AI personal shoppers that curate a selection for consumer approval. Smart home systems that suggest and pre-approve routine replenishment orders pending one-click human confirmation.

This is where Agentic Commerce begins. At Level 2, the AI agent has assumed the search and evaluation function. The consumer is no longer browsing. The consumer is reviewing the agent's work. The Shopper Schism® has begun: the consumer and the shopper are no longer the same entity. The consumer experiences value. The agent executes the shopper function — with a human checkpoint.

Commercial strategy: Both GEO and AIO begin to matter. GEO determines whether your product enters the agent's consideration set (the shortlist presented to the consumer). AIO determines whether the agent ranks your product at the top of that shortlist. At Level 2, the human still has veto power, so brand recognition provides a safety net. But the agent's ranking is the primary determinant of selection. Products presented first are purchased most.

The trajectory: The most dynamic zone on the spectrum. Level 2 is where mainstream consumer categories will sit for the next two to four years. The competitive stakes here are enormous: the agent's shortlist is the new shelf space.

Level 3: AI-Managed Commerce

What it looks like: The consumer sets standing preferences and rules. The AI agent executes purchasing decisions autonomously within those parameters. The consumer is notified after the purchase, not before. The consumer can review and return, but did not participate in the decision.

Examples today: Smart refrigerators that automatically reorder milk when inventory drops below a threshold. Subscription management services that optimise across competing providers without human intervention. B2B procurement systems that auto-purchase standard supplies within pre-approved budgets and vendor lists.

The Shopper Schism® is complete at Level 3. The consumer and the shopper are fully separated. The consumer drinks the milk, wears the clothes, uses the supplies. The agent bought them. The consumer may not know which brand the agent selected, which alternatives it considered, or what trade-offs it computed. The consumer delegated not just search and comparison but judgement.

Commercial strategy: AIO dominates. The human is not in the decision loop. Brand recognition is irrelevant at the moment of purchase — the consumer does not see the options. The competitive battle is entirely algorithmic: structured data quality, API accessibility, delivery reliability, pricing transparency, and computational trust signals determine selection. The Four Ds Framework™ (Data Quality, Discoverability, Decisional Clarity, Delivery Reliability) is the competitive toolkit.

The trajectory: Expanding rapidly in commodity categories and routine purchases. Consumer electronics replenishment, household consumables, basic personal care, office supplies, and standardised B2B inputs are migrating to Level 3 now.

Level 4: AI-Autonomous Commerce

What it looks like: The AI agent operates with broad strategic mandates rather than specific purchasing instructions. "Manage my household budget and keep everything stocked." "Optimise my company's procurement for cost and sustainability." The agent makes purchasing decisions that the consumer may never individually review. The agent manages the consumer's commercial life as a delegated portfolio.

Examples emerging: Multi-category AI agents that manage entire consumption portfolios. Corporate procurement AI that autonomously manages supplier relationships, renegotiates contracts, and switches vendors based on performance data. Agent-to-agent commerce where purchasing agents interact directly with vendor agents, negotiating and transacting without human involvement at any point.

This is the terminal state of the Spectrum. At Level 4, commerce operates in what Accornero terms the Automaton Economy™: algorithms transacting with algorithms, with human oversight reduced to exception handling. The consumer sets strategic intent ("keep me healthy, keep costs reasonable, prefer sustainable options"). The agent translates intent into thousands of individual purchasing decisions across categories, vendors, and time horizons.

Commercial strategy: AIO at its most demanding. Layer 3 of the Agent Decision Preference Stack is the entire battlefield. Products must be discoverable, evaluable, and transactable through purely machine-to-machine interfaces. Human-facing marketing is not merely insufficient; for Level 4 transactions, there is no human to face. The competitive advantage belongs to organisations with the most complete, most accurate, most reliable operational infrastructure.

The trajectory: Nascent but inevitable. Level 4 is where B2B commerce is heading fastest (procurement automation) and where consumer commerce will follow as agent trust is established. The governance challenges at Level 4 — algorithmic collusion, the Shadow Principal Problem, liability allocation — are the subject of the Governance Gauntlet™ framework.

The Spectrum at a Glance

Level	Name	Human Role	AI Role	Where Marketing Competes
0	Human-Only	Full control	None	Traditional channels
1	AI-Informed	Decision-maker	Advisor	GEO (citation and visibility)
2	AI-Guided	Final approver	Shortlist curator	GEO + AIO (consideration + ranking)
3	AI-Managed	Retrospective reviewer	Autonomous buyer (within rules)	AIO (operational performance)
4	AI-Autonomous	Strategic intent setter	Portfolio manager	AIO at machine-to-machine level

Where Is Your Category?

The strategic question is not whether your category will migrate up the Spectrum. It is how fast, and whether you will be ready when it does.

Commodity consumer goods (paper towels, batteries, basic toiletries) are already at Level 2–3 for early adopters and heading toward Level 3–4 within three years. Consumer electronics are at Level 1–2, migrating to Level 2–3. Complex B2B services remain at Level 0–1 but are being disrupted at the procurement layer. Luxury and high-involvement purchases will remain at Level 0–1 the longest — but even here, AI research assistance is pushing the informational boundary upward.

The Spectrum of Delegation is not a prediction. It is a diagnostic. Plot your category. Plot your competitors. Then ask: is my organisation ready for the level my category is moving toward?

The Algorithmic Readiness™ Assessment provides a structured diagnostic to answer that question.

Frequently Asked Questions

At what level does traditional marketing stop working?

Traditional marketing remains fully effective at Level 0 and partially effective at Level 1 (where the human still makes the final decision after AI research). At Level 2, marketing influences whether the human approves the agent's recommendation but does not determine the recommendation itself. At Levels 3 and 4, traditional marketing is structurally irrelevant to the transaction.

Can a category move backward on the Spectrum?

In theory, yes — a major trust breach (Phase 3 of the Trust Paradox™) could cause consumers to revoke delegation and return to lower levels. In practice, the trajectory is one-directional. Convenience, time savings, and superior outcomes at higher delegation levels create powerful retention effects. Consumers may pause their migration but rarely reverse it permanently.

Does every product need to be ready for Level 4?

No. The investment should track the migration speed of your specific category. A luxury watchmaker does not need Level 4 readiness today. A consumer electronics brand does. The Spectrum is a planning tool, not a prescription. Match your investment to your category's trajectory.

Cite This Page

Accornero, P.F. (2026). "The Spectrum of Delegation: Five Levels of AI Commerce." The AI Praxis. Retrieved from https://www.theaipraxis.com/spectrum-of-delegation