The old buying cycle was notice, click, reveal intent. The new one starts with a person stating an intent to an agent, which researches, narrows the options and comes back with a recommendation. Being seen matters less. Being chosen is everything.
This article explains how that cycle works, what an agent returns when asked a simple question, how it weighs risk, and the framework Monarchy uses to think about it, the Selection Stack. It closes with what changes when the interface moves beyond the screen. Opinions are labelled as Monarchy’s view or Andrea d’Agostini’s view.
01From Attention to Selection: How Agents Choose
The buying cycle used to be: a person notices something, clicks, and behaves in a way that reveals intent. The new cycle is: a person states an intent to an agent, the agent researches and narrows the options, the person approves, and the agent completes the task.
Source: Monarchy, from the cycle described above
Consumer willingness is already meaningful. In Salesforce research published in April 2025, 24% of consumers said they were already comfortable with AI agents shopping for them, rising to 32% among Gen Z; 39% were comfortable with agents scheduling appointments for them1.
Source: Salesforce 1
What an agent returns
In our lab, we gave a generic request to the major AI platforms: build a podcast studio, high quality, without overspending. Each came back with one camera, one microphone and one setup, an explanation of why, and an offer to buy them. No ads. No photos. Just a recommendation and a question. (Monarchy internal test; illustrative, not a statistical study.)
A simple behavioural test shows the same shift. Take any product or service: a person asks an assistant for the best option within a budget, with the conditions that matter to them. The agent looks at thousands of options and returns three. Who saw the ads for all the others? Nobody.
How agents decide
An agent is not impressed. It is confident or it is not. It knows the person’s history, budget and constraints, and it looks for the option that fits best and carries the least risk. For a machine, risk looks like inconsistency: a price that differs between your site and a marketplace, a warranty that changes by retailer, reviews that contradict your claims.
↓ The agent reads down through the stack
The human moment
When the agent hands the client to you, the call, the visit and the conversation must confirm what was promised. Voice, tone and care become the moment of truth.
Preference
Brand, trust and community, built before the question is asked. The strongest position is when the buyer tells the agent, “find me the best one, and check whether it’s them”.
Verifiable proof
Reviews, case studies with real outcomes, awards and partners who vouch for you, in places you do not control. Anything you say about yourself counts for less every year.
Consistency
The same facts everywhere you appear: your site, marketplaces, directories, review platforms and social profiles. Agents read inconsistency as risk.
Structured truth
What you do, for whom, at what price or pricing logic, what is included and what is guaranteed. Written once, clearly, in a machine-readable way.
Beyond the screen
The interface is moving too. Agents increasingly work through voice, earbuds and glasses as well as phones and laptops. When the agent answers out loud, there is no page of results to buy your way onto. There is one sentence about you.
Download the PDF. Give it to your agent.
Have your agent study our white papers alongside you.
Free download
Want the full version?
Leave your email and we’ll send it.
Thanks, you’re on the list.
We’ll send it to .
Adapted from the Monarchy white paper “You are not talking to a human” (2026) by Andrea d’Agostini. Sources are listed below. Statements marked as views are opinions.
Sources
- Salesforce, “Consumers Are Ready for AI Agents. Are Businesses?” 14 April 2025 (24% comfortable with agents shopping for them; 32% of Gen Z; 39% with agents scheduling appointments). https://www.salesforce.com/news/stories/consumers-ready-for-ai-agents-research/
