While the holidays are still on our minds (even if, for some, they are already a distant memory), we'd like to reflect on what our journey looked like this summer. We don't mean the holiday itself, but the customer journey that led us to it: how did we choose our destination this year? How did we plan our route once we got there? What about accommodation and transport? And how did we pay for it all? 

For many of us, that dream trip probably began with a simple prompt to our go-to LLM: “We're a family of four with two teenagers and we'd like a mountain destination that isn't too hot and isn't too expensive…” From there, we dived into its suggestions and kept the conversation going over several iterations, without a traditional Google search and with hardly a click.

The most tech-savvy travellers may have turned to their own personal travel assistant, an AI agent built and trained on their profile, preferences, travel history, budget limits and other requirements.

What we experienced this summer as consumers, handing the planning of our trip over to an LLM or an AI agent, is exactly what is starting to happen with the way your customers buy and decide.

The arrival of LLMs in decision-making, in travel and in countless other sectors, has had and continues to have many implications for brands. But the emergence of agentic AI systems opens up a new scenario with plenty of loose ends still to be understood.

In this article we share 5 market signals that we at RocaSalvatella have analyzed and that we believe point to where the so-called “agentic economy” is heading.

Shall we take a look?

1. A new intermediary between brand and customer

The agent is a new type of brand stakeholder, with a new profile and its own codes of engagement. It is rational, data-driven, has no brand loyalty and is immune to storytelling: the agent only evaluates structured attributes. 

A traditional distributor sits “between” customer and brand. A prescriber exerts influence “between” customer and brand. A marketplace organises access “between” the two. An AI agent, however, can act as the customer's representative in dealings with the brand. A representative to whom “powers can be delegated” so that it makes decisions on the customer's behalf.

So what does this mean for brands? To be visible and lead in share of model (the new share of voice), you need to be agent-readable: open APIs, organised data, machine-readable product catalogues, quality content and positive reviews… A scenario that is changing communication and marketing strategies, and one we explored in depth a few months ago in our GEO strategy guide.

2. The intention economy gains ground on the attention economy

Some authors argue that the agentic economy revolves around capturing consumer intention. This runs counter to the model that has dominated digital business until now, the attention economy, in which brands compete to attract consumers and generate clicks, impressions, likes, screen time…

Much to the dismay of the author who coined the original term Intention Economy back in 2006, envisaging an optimistic future in which empowered consumers, owners of their personal information, freely chose which providers to share their data with, the new spin on Intention Economy in an agentic scenario is somewhat less aspirational. 

In the new intention economy, AI agents compete to understand the user's purpose, preferences, and context with the sole aim of giving a relevant answer that solves their problem. Capital One illustrates this well in its explanation of MACAW, its agentic architecture, and the foundation of Chat Concierge, the company's first agentic system to interact directly with customers.

So what does this mean for brands? This new intention economy disrupts the current advertising model. Adapting to the new dynamics means combining traditional digital marketing strategy with agent strategies that detect consumer intention in order to tailor and deliver a personalized, relevant proposition in real time. We face the challenge of managing a new “dual relationship model," as an article in California Management Review (Berkeley) calls it: one relationship with people and another with algorithms.

3. Agents search, select and steer the decision: the end of “traditional browsing”? 

As consumers, we often delegate decisions to AI agents in two situations: when the complexity exceeds our human capacity to process information or reason it through (e.g. exhaustive comparisons of prices, terms and alternatives), and when the value of the decision does not justify the effort or follow-up needed to do it any other way.

Salesforce predicts that during the next Cyber Week 20% of all ecommerce traffic will come from AI chat agents, and that 1 in 3 ecommerce sites will have its own personalised shopping agent up and running. Adobe, for its part, reports that LLM-generated traffic converts 31% better than all other sources combined (paid search, organic search, social media, etc.). 

This is happening not only in B2C markets but in B2B markets too. Procurement departments, as in the case of BMW with its AIconic system, in operation since 2025, are increasingly searching for, evaluating and contracting suppliers through agentic AI processes. Gartner estimates that by 2028, 90% of B2B purchases will be intermediated by AI agents and that agent-managed spend will reach $15 trillion globally.

So what does this mean for brands? Beyond adapting our marketing strategy to become visible to new agentic buyers, in this new intermediated market, commercial and customer relationship models are being rethought. Sales and customer service teams are asking questions such as: for which markets, products or touchpoints do we need a person to provide advice and manage the commercial relationship? Which sales tasks and information can AI agents handle? What do we gain, and what risks do we run, if we let AI agents interact directly with our customers?

The experience of those who have already redesigned their sales teams around agentic AI points to a clear division of labour: agents take on administrative coordination (prospecting, drafting proposals, CRM), while the human role focuses on executive relationships and complex negotiations.

4. Transactional agents: an ungoverned market?

Unlike LLM-based assistants, which tell us what to do and how to do it, agents can decide autonomously, with no human intervention, what to buy and from whom, when to buy and what price to pay. 

We are closer than we think to the moment when our personal AI travel agent buys our plane tickets for us as soon as the fare drops, even while we sleep, as a Mastercard executive explained in a video a few months ago, following the launch of the company's agentic payments protocol.

Agentic payments take us into a new scenario: one in which the entire funnel (search, consideration, decision and purchase) takes place no longer just in a single channel but in a single environment: the one in which the agent operates.

So what does this mean for brands? Agents' ability to transact and to interact with other agents autonomously raises a host of unresolved dilemmas: are these transactions secure? Who is responsible for any security breaches or errors in them? 

Transactional agents also call into question the very purpose of app-based ecosystems since, as we have just seen, users buy directly without leaving their LLM environment. They also open a debate about the risk of handing the end-to-end customer relationship over to third parties.

5. Reputation and trust are shifting

Agents are not only emerging as new intermediaries sitting, more or less uncomfortably, in the middle of the brand-customer relationship. They are often taking over trust and reputation as well.

The trust barometer carried out by the Edelman Trust Institute in the US, China, Brazil, the UK and Germany shows that 67% of consumers who trust AI would trust it to manage their finances. 

Let's be honest: who do we trust more? A human salesperson working for company X, or an LLM that “objectively” compares several market alternatives based on the criteria and preferences we give it? 

So what does this mean for brands? The challenge is clear: how can a brand that has invested years and resources in building its reputation and earning customers' trust compete with an agent perceived as a neutral, “altruistic” guardian of the customer's interests?

Some will think the best strategy is to sit by the river and “wait for the body of your enemy to float by”. In other words, to wait until AI agents start showing signs of being sponsored by certain brands and their supposed neutrality is called into question. 

Either way, it is clear that trust is not an added value: it is an essential condition for the agentic market at scale. And yet the legal and reputational responsibility for agentic decisions, or for decisions shaped by agents, still has no clear owner.

Tips and key takeaways:

1. POSITIONING your brand in AI environments matters more and more→ Audit today whether your brand is visible in AI and what AI says about it when a customer asks, and fix it before your competitors do

2. AI environments are defining a new INTENTION economy→ Train your agentic systems properly so they can capture the customer's context and job to be done, and offer a personalised, relevant proposition that solves their problem and ties them to your brand.

3. A new key capability emerges: being ATTRACTIVE AND ELIGIBLE for an AI agent→ Turn catalogues, demos and content into structured, verifiable data so that agents include you in their consideration set and recommend you.

4. AI agents are the NEW PURCHASING INTERMEDIARIES→ Redesign your commercial model with potential agentic buyers in mind, and assess how your own AI agents, working in hybrid internal teams, can reinforce your human sales team and extend what you can do together. Weigh up the benefits and risks of letting customers complete their purchase in environments outside your ecosystem.

5. TRUST is moving to AI environments. → Define your strategy to compete and/or collaborate with third-party agents that are capturing customer trust. Build trust in the new environment through content, context and authority that shape how these agents read you.