How should B2B suppliers prepare for agentic commerce?

McKinsey QuantumBlack published the agentic commerce automation curve in January 2026, projecting AI agents will mediate $3 trillion to $5 trillion of global consumer commerce by 2030. The framework describes six levels of buyer delegation. This article applies the supplier-side lens: at each level, what does a B2B business need to be discoverable, trusted and transactable enough to be cited, recommended and selected by AI agents acting on behalf of buyers.

By Helenor Rogers, CEO of KnownEntity.ai

McKinsey's most strategically important piece of 2026 so far is not about AI productivity. It is about AI buying. Published 28 January 2026 by McKinsey QuantumBlack, The automation curve in agentic commerce projects that AI agents will mediate $3 trillion to $5 trillion of global consumer commerce by 2030 even under moderate scenarios. The framework lays out six distinct levels of delegation, from rules-based replenishment to fully networked multiagent marketplaces.

The framework is rigorous. The question it leaves unanswered is the one suppliers need to address right now: what does it take to be in the answer when the agent arrives?

McKinsey describes how much buyers will delegate to agents. It does not describe what suppliers need to do to be selectable by those agents at any level of the curve. That is the supplier-side question, and for many B2B businesses and brands, the answer in May 2026 is: not yet enough. McKinsey's own framing on this point is direct: if your catalogue, policies and value proposition are not machine-readable, agents and the shoppers who delegate to them simply will not find you, no matter how beloved your brand is. That is the entry ticket, everything above the entry ticket is competitive advantage.

So now to some clear definitions. The McKinsey automation curve is a six-level framework describing how much of the buying journey buyers will delegate to AI agents, from rules-based replenishment at Level 0 to fully networked agent-to-agent commerce at Level 5. Supplier-side legibility is the practice of ensuring that at each level of delegation a buyer reaches, the supplier remains discoverable, trusted and transactable to the AI agent acting on the buyer's behalf. AI Authority Infrastructure (AIAI) is our supplier-side response to the buyer-side curve. At KnownEntity.ai, we use our proprietary AI Authority Ladder™ methodology to build that capability across four stages:

1. Diagnostic — how AI currently sees your business across all large language models e.g chat GPT.
2. EntityCore™ — our proprietary knowledge graph build, the structured machine readable content that is your companies source of truth, that AI agents can discover, trust and transact with.
3. Citation strategy — building third-party references that AI trusts.
4. Ongoing measurement — tracking citation performance across platforms and updating the EntityCore™.

Here is what the McKinsey curve looks like, what it requires of suppliers at each level, and what it means specifically for B2B businesses and brands.

The McKinsey six-level automation curve in plain terms

McKinsey's curve runs from Level 0 to Level 5:

McKinsey's curve runs from Level 0 to Level 5:

• Level 0, programmed convenience. Subscriptions and scheduled refills, the pre-agentic baseline.

• Level 1, assist. The agent researches, compares and synthesises but the human still buys.

• Level 2, assemble. The agent builds a purchase-ready basket and the human approves.

• Level 3, authorise. The human pre-defines rules and the agent executes within them, escalating only on exceptions.

• Level 4, autonomise. The agent operates against standing goals such as keeping household essentials under £300 a month or maintaining airline loyalty status at the lowest cost.

• Level 5, networked autonomy. Agent-to-agent commerce running continuously in the background, with personal agents negotiating directly with specialised merchant agents, logistics agents, payment agents and loyalty agents.

The crucial McKinsey insight is that delegation does not move uniformly up the curve. It accelerates in low-regret, high-repetition categories like consumables, plateaus in identity-driven categories like luxury goods, and is selective in complex purchases like travel and home electronics. Different categories settle at different ceilings. The aim is not maximum autonomy but optimal delegation.

For B2B, McKinsey's sidebar is unambiguous. The same curve applies, but the stakes change. In B2B, delegation is institutional rather than personal. Authority flows from procurement policies, budget owners, risk teams and legal frameworks. Once unlocked, it scales far more powerfully than in consumer settings. The real B2B inflection point is supervised execution: when agents are authorised to act within clearly defined policies such as spending thresholds, preferred vendors and compliance rules, they can manage replenishment, renewals, substitutions and exceptions automatically.

That is the world unfolding right now in procurement, packaging, materials, professional services renewals and software licensing. It is not theoretical. It is being built into existing eProcurement systems, Shopify Catalog, Google's Universal Commerce Protocol and the open agent protocols (MCP, A2A, AP2, ACP, UCP) that the Linux Foundation's new Agentic AI Foundation is now stewarding.

The supplier-side curve: what businesses need at each level

McKinsey's curve is buyer-facing. To make it commercially actionable, suppliers need to read it backwards. At each level of buyer delegation, the supplier requirements escalate. Discoverability gets you onto the entry ramp. Trust keeps you in the consideration set. Transactability is what makes you selectable when the agent is empowered to execute.

• At Levels 0 and 1, discoverability is the floor. The agent is researching and comparing on the buyer's behalf. If your business does not appear in the AI's synthesis, you are not in the conversation. McKinsey is explicit here: verifiable data beats marketing gloss. Agents need structured attributes, clear eligibility rules, sizing and fit certainty, and claims that can be substantiated.

• At Level 2, trust becomes load-bearing. The agent is now assembling a purchase-ready basket and resolving trade-offs. McKinsey's prescription is API-first merchandising: inventory, pricing, shipping promises, promotions and returns logic exposed cleanly. Suppliers whose data is fragmented across PDFs, sales decks and human-only contact pages are systematically deprioritised at Level 2. The agent simply chooses suppliers it can model with confidence.

• At Level 3, transactability starts mattering commercially. The buyer has authorised the agent to execute within rules. McKinsey's requirements for merchants are precise: purchasing authorisation that can be limited (by budget, time window, merchant or category), activity that can be audited, and actions that can be reversed. Suppliers without these are not eligible for the agent's shortlist, regardless of historical relationships.

• At Level 4, the competitive logic shifts entirely. McKinsey's framing is sharp here: competition shifts from winning a single purchase to earning a place in the agent's ongoing plan. Merchants need deeper integration around loyalty, eligibility, substitutions and service guarantees. Put plainly, it is no longer enough to expose a catalogue. Suppliers must expose the rules and policies that determine what good looks like.

• At Level 5, networked autonomy, the world becomes agent-to-agent. Personal and procurement agents negotiate directly with specialised merchant, logistics, payment and loyalty agents. Trust is carried through reputation signals. Transactions settle through shared protocols. McKinsey's warning is direct: those who do not expose policies, guarantees and loyalty logic in machine-readable ways risk becoming interchangeable suppliers competing primarily on price in machine-negotiated flows.

Let's look at this across some key sectors:

What this means for food and drink manufacturing

Food and drink manufacturing sits at the high-velocity end of the curve. Replenishment, packaging, ingredient sourcing and contract manufacturing fall squarely into the low-regret, high-repetition categories where McKinsey expects delegation to accelerate fastest. Buyers in retailer procurement, food service operators and contract caterers will increasingly delegate ongoing supplier decisions to agents working within procurement policies.

Three things will determine which manufacturers are selected. Structured product, certification and capacity data that agents can read directly. Third-party citations from trade press, analyst coverage and customer references that confirm the supplier's standing in the category. Transactable interfaces such as API-exposed pricing, lead times and reorder logic. Manufacturers that have all three become default suppliers. Those without them are routed around.

What this means for construction

Construction sits closer to McKinsey's selective delegation territory. Specification-driven purchases, regulatory compliance, project-specific design requirements and the involvement of architects, engineers and contractors keep the human in the loop on the major commitments. But the long tail of materials procurement, fixtures, fittings, mechanical and electrical components and ongoing site supplies is moving towards Levels 2 and 3 quickly.

The supplier-side requirement for construction businesses is to be specifically machine-legible at the technical attribute level: dimensions, performance ratings, compliance certifications, sustainability credentials and compatibility data. AI agents researching on behalf of project managers and quantity surveyors will narrow the shortlist before any human contact. Construction suppliers without complete, structured technical data are increasingly invisible at the longlist stage.

What this means for professional services

Professional services is where McKinsey's curve gets most interesting strategically. The headline professional services purchase, choosing a strategy consultancy, a law firm or an audit partner for a major engagement, sits high on the identity and regret-risk axis. Delegation plateaus low. Humans stay in control. McKinsey's framing applies: the agent's role is to inform deliberation by exposing rich contextual attributes, provenance and long-term value signals while preserving human control at the point of commitment.

Professional services also has a substantial Level 1 to Level 3 underbelly: ongoing advisory retainers, compliance services, fractional CMO engagements, accountancy and tax services, IT managed services, recruitment and other professional functions where the buyer is willing to delegate research, comparison and increasingly, supervised renewal. The firms that win at the deliberative end are the firms whose expertise, methodology and named partners show up in the AI's research synthesis. The firms that win at the recurring end are those with structured service definitions, clear scope rules and transactable engagement models.

In both cases, the supplier requirement is the same: be present in the AI's answer, accurately represented, with enough machine-readable substance behind the brand for the agent to do its job confidently. You can read about how AI agents build that understanding of your business in our white paper The Invisible Shortlist.

The strategic point McKinsey leaves implicit

McKinsey's curve assumes the supplier is in the answer. For most B2B businesses in May 2026, that assumption is not yet justified. The agentic commerce era will not be won by the biggest brand or the loudest marketing. It will be won by the suppliers AI agents can find, trust and transact with at every level of delegation their buyers choose to operate at.

The window for getting this right is narrower than it looks. McKinsey's projection of $3 trillion to $5 trillion of agent-mediated commerce by 2030 implies a build-out happening now, not in 2030.

Suppliers that establish discoverability, trust and transactability in 2026 will compound authority advantage across the curve. Suppliers that wait for the curve to stabilise will find themselves competing on price in machine-negotiated flows, exactly the outcome McKinsey warns against at Level 5.

How to start building AI Authority for your business

KnownEntity.ai is a UK-based AI Authority Infrastructure consultancy that works with B2B and branded organisations to ensure they are discoverable, trusted and transactable by the AI agents their buyers now use. We work with food and drink manufacturers, construction firms and professional services businesses. We use our proprietary AI Authority Ladder™ methodology to take businesses from invisible to appropriately and strategically cited, step by step and ready to respond to buying agents.

"Most businesses we speak to assume the goal is simply to appear in AI answers. But appearing for the wrong things, or in the wrong context, can be just as damaging as not appearing at all. The real work is ensuring AI agents understand what your business genuinely stands for, and cite you for the things that matter to your commercial strategy. That's what we mean by being properly known."

— Helenor Rogers, CEO of KnownEntity.ai

If you would like to understand where your business currently stands in AI search, and what it would take to close the gap, get in touch at hello@knownentity.ai or visit knownentity.ai.

Last updated: 10 May 2026

Sources referenced in this article:

• McKinsey & Company QuantumBlack, "The automation curve in agentic commerce", Deepa Mahajan, Hannah Mayer, Katharina Schumacher, Roger Roberts and Katharina Giebel, published 28 January 2026

• McKinsey & Company QuantumBlack, "The agentic commerce opportunity: How AI agents are ushering in a new era for consumers and merchants" (17 October 2025), source for the $3 trillion to $5 trillion 2030 projection

• Linux Foundation, announcement of the Agentic AI Foundation, anchored by founding contributions including Model Context Protocol (MCP), Goose and AGENTS.md (December 2025)