What happens to advertising when agents don't care about brands

I. The end of advertising as we know it

The global advertising market was worth about $876 billion when I first wrote this, in January 2026, with forecasts of $2.5 trillion by 2032. Most of it is spent convincing humans to choose one brand over another.

What happens when your customer is no longer human?

What happens when AI agents handle purchasing decisions, comparing options on verified data instead of emotion? When a billion-dollar brand means nothing to an algorithm that only cares about reliability scores, delivery windows and price-performance ratios?

The advertising industry is built on one assumption: humans make irrational decisions, and those decisions can be influenced. Remove that assumption, and the entire edifice crumbles. What gets built in its place is infrastructure for what comes after advertising.

II. The intention economy

Today's economy runs on attention. Capture eyeballs, manufacture desire, convert impressions into purchases.

Tomorrow's runs on intention. Broadcast what you need, receive what matches, transact.

On the demand side, a human or their agent broadcasts an intent:

[INTENT: DEMAND]
Need:         Running shoes
Use case:     Marathon training
Budget:       €120-180
Delivery:     By Friday
Constraints:  reliability score > 95%
              sustainability rating > B
              no animal products
Weights:      60% durability, 30% comfort, 10% aesthetics

The intent enters the network, open infrastructure where specialised matching agents operate, and no single platform owns it.

On the supply side, producers and sellers, or their agents, broadcast capabilities:

[INTENT: SUPPLY]
Product:      EnduraRun Pro X
Category:     Running shoes / marathon
Price:        €149
Delivery:     2 days (EU), 4 days (global)
Verified:     reliability 97.3% (12,847 verified purchases)
              sustainability A- (certified supply chain)
              materials: synthetic, vegan
              durability 8.7/10 (prediction-market consensus)

Matching agents then read both sides. They optimise for match quality, how well a given supply satisfies a given demand, with ad spend and click-through rates out of the equation. The buyer's agent receives ranked options because the match is better, never because a brand paid more.

III. De-brandization

Brands exist because trust is expensive to verify.

When you buy a famous pair of running shoes, much of what you pay for is a trust shortcut. The company spent billions building it, teaching you that the logo means quality, that the ads mean status, that the presence means reliability.

But what if trust becomes cheap to verify?

A brand does three jobs, and each one changes differently.

Trust signal. Today: "I trust their quality because they are a big brand." Tomorrow: "this product has a 97.3% reliability score across 12,847 verified purchases, certified at the source."

Search shortcut. Today: "just get me the shoes from that brand." Tomorrow: "get me running shoes matching these specifications", followed by the parameters.

Identity signal. Today: "I am that kind of person." This one survives, because humans still want identity markers. It becomes one factor among many, and agents have no identity preferences at all.

When agents make purchasing decisions, or merely filter options for humans, brand spending becomes inefficient. You cannot emotionally manipulate an algorithm. What counts instead is reliability measured on purchase outcomes, attributes certified along the supply chain, quality scores formed by people who staked something on them, and real-time availability. The supplier who delivers consistent quality wins over the supplier with the biggest budget. That is de-brandization: the move from brand-as-trust-proxy to verification-as-trust-infrastructure.

IV. What the infrastructure needs

Five components make intentions meet.

An open intent network. Demand and supply intents are broadcast to infrastructure any agent can join, with no platform extracting rent in the middle. Every intent is structured, queryable and composable.

Specialised matching agents. Category specialists, constraint optimisers, logistics coordinators, quality verifiers. They compete to produce the best matches, earn fees when a match succeeds, and lose reputation when it fails.

A verification layer. Every claim can be checked: attributes attested at the source, supplier reliability computed from verified outcomes, delivery accuracy tracked and scored.

Prediction markets. What is the real durability of this product? Will this supplier deliver on time? Is this sustainability claim accurate? Each question can become a market where participants stake value on the outcome. Reviews can be gamed; a market where being wrong costs money is much harder to game, and it becomes the reputation infrastructure of the intention economy.

Settlement. When a match executes, payment moves through escrow, released on delivery, and the outcome feeds back into the reputation scores. Everything stays auditable.

V. Three kinds of agent

Consumer agents know your preferences, constraints and history. You express an intent in plain language, or the agent infers it from context; it turns that into a structured intent, broadcasts it, filters the matches, shows you options or acts within your parameters, handles the negotiation and the transaction, and learns from the result. You do not browse, you do not comparison-shop, you do not see ads.

Supplier agents watch inventory and capabilities in real time, broadcast accurate supply intents, answer matching queries, negotiate within authorised parameters, and update the attributes as outcomes arrive. The supplier's job moves from marketing to delivering verifiable quality; the agent handles discovery.

Matching agents are neutral infrastructure: they index intents on both sides, rank by match quality, broker the negotiation, earn fees for successful matches and build reputation on the satisfaction they produce.

This is what I mean by a multi-species economy: humans express intentions, agents execute and coordinate, contracts enforce agreements, markets surface truth, and verification replaces faith. Humans and AI working as economic peers, with AI as a participant in coordination rather than a tool held by one side of it.

VI. What changes

For buyers, the change is direct. Today you are bombarded with ads while brands compete for your attention, and you decide on incomplete information and emotional pressure. Tomorrow you state an intent, your agent finds matches, and you see options ranked by how well they meet your actual needs.

For suppliers, the economics change at the root. Many spend a large share of revenue on marketing and compete for attention as much as on quality; brand building can matter more than product building. In the intention economy you focus on verifiable quality, your agent broadcasts your capabilities, and customers find you through match quality. The marketing budget becomes a product budget.

For the economy as a whole, the resources now devoted to capturing attention get redirected. Brands work as rent-seeking intermediaries between quality and customers; when verification is cheap, resources flow to actual value creation, coordination costs drop, quality wins, and manipulation becomes economically irrational.

VII. The transition

The intention economy will not arrive overnight, and I wrote down three phases for it in January 2026. A hybrid period, through 2027, where intent matching runs alongside traditional advertising for early adopters and selected categories, and reputation data starts to accumulate. An agent-assisted period, roughly 2027 to 2029, where agents handle routine purchasing and intent matching becomes the default in some categories while brand importance declines in agent-mediated transactions. And an agent-native period from about 2029, where most commercial transactions are agent-mediated, and brands persist for identity and luxury while fading for commodities.

Infrastructure for the third phase has to be built during the first: intent protocols that work for humans today and agents tomorrow, matching that scales from manual to autonomous, verification that accumulates trust before it is critical. The infrastructure needs to exist before the agents fully arrive.

The company building in this space inside the Fucina Nexus Program is Mercatalia, whose function is connecting supply and demand on multi-species intents. It is in its design phase, with its first functional proof planned for November 2026: a structured intent, constrained offers, disclosed sponsorship, and a recorded outcome. That is all I will say about it here.

VIII. Where the money goes

Hundreds of billions flow through advertising every year. Most of it pays for persuasion, attention capture and brand building as rent.

When agents stop responding to persuasion, when verification replaces trust and intent matching replaces advertising, the money keeps flowing, along different lines: from manipulation budgets to quality, from brand building to capability building, from capturing attention to creating value. That flow is what comes after advertising, and it is worth building the rails for now.