A framework for what happens when AI stops reacting to what you've done, and starts anticipating what you're about to want.
For fifteen years, marketing has run on one principle: if you did X, you'll probably want Y. That's a recommender system. It's reactive by design. It sells to the person you were five minutes ago, not the person you're becoming.
Latent Intent Marketing is about the next shift: AI agents that notice a pattern forming across scattered signals — a purchase here, a search there, a gap opening in your calendar — and infer where you're heading before you've said so yourself. Not mind-reading. Probability.
The series runs in five parts. Each one builds on the last.
Part 1 — Latent Intent Marketing
When machines model trajectory, not just behaviour
The founding piece. Lays out the core distinction between reacting to behaviour and modelling trajectory, and introduces the framework's two supporting ideas: a Real-Time Agent Marketplace (RTAM), where both buyer and seller are represented by agents, and Agent Engine Optimisation (AEO), the successor to SEO for a world where machines do the shortlisting. The clearest illustration is a swarm of personal agents noticing you're edging toward a half marathon in Cyprus, and assembling the whole trip before you've consciously decided to enter the race.
Part 2 — The Architecture of Agent-Mediated Markets
The Six-Layer Stack
If agents are going to sit between brands and customers, what actually holds that up? Part 2 breaks the system into six layers: signal capture, intent modelling, agent orchestration, marketplace protocol, brand interface, and the objective function sitting quietly on top, deciding what the whole system optimises for. Whoever controls a given layer controls a different kind of leverage, and the piece closes by setting up three possible futures for where that control settles.
Part 3 — If Agents Filter What We Buy, What Happens to Advertising?
Scenario One: what happens to the old order
Two decades of advertising, performance marketing, branding, and agencies were built to win human attention. This part asks what happens to that machinery once filtering happens upstream, before a human ever sees the options. The short version: none of it disappears, but all of it mutates. Attention stops being the scarce resource. Selection is. The question shifts from "did you get seen" to "did you survive the agent's shortlist" — and that changes what performance marketing, branding, and agencies are actually for.
Part 4 — The Model-Centric Future
Ad Wars: A New Beginning
If agents mediate markets, models mediate agents. This part looks at what happens when a small number of AI labs become the reasoning layer underneath every agent, shopping, travel, health, finance alike. The argument: this kind of concentration isn't imposed from outside, it's reinforced by how these systems actually improve. Monetisation shifts from buying attention to influencing inference. The next commercial fight isn't over impressions. It's over inference.
Part 5 — The Opportunity Layer
Where value moves next
Every infrastructural shift displaces some work and creates other work. This is the practical payoff of the series: where the opportunity actually sits once agents start mediating markets. Covers Agent Engine Optimisation as a service category, the infrastructure layer underneath agent marketplaces, products built to be legible to machines rather than humans, and where data becomes durable competitive leverage rather than just analytics fuel.
Latent Intent Marketing
is an ongoing framework, developed publicly. Get in touch if you'd like to talk through how it applies to your product or category.