Two years ago we were talking about generative AI as another tool in the marketer's belt. Today the conversation is different. The industry is no longer discussing whether artificial intelligence writes copies or segments audiences; it is discussing whether it will be an autonomous agent that launches the next campaign without anyone pressing “Publish”. This change is being called agentic marketing, The data published in recent months make it clear that this is not a technology trade fair slogan.
Next week, from 9 to 11 June, Malaga is hosting DES 2026 - Digital Enterprise Show - with artificial intelligence applied to marketing as one of its central themes, and a space exclusively dedicated, Digital Marketing Planet, to these innovations. It's a good excuse to take a break and sort out what's really going on with AI agents in marketing, what the analysts are saying and what you should be looking at if you manage a digital budget in an SME or a big brand.
What do we mean by “agentic marketing”?”
An AI agent is not an assistant waiting for your instructions. It is a system that receives a goal - for example, “increase landing X conversion by 15%” - plans a sequence of actions, executes them on multiple platforms, measures the outcome, and adjusts its approach without asking for human approval at each intermediate step. That autonomy is the key difference from classic marketing automation, where one person was still driving every strategic decision.
In practice, an agent can review behavioural signals in your CRM, identify a segment of users at risk of churn, compose a different email for each subgroup, schedule it, decide the optimal day and time per person and, once sent, reallocate budget in Meta Ads or Google Ads based on the observed return. It does this in a matter of minutes, not weeks.
What analysts say: pace of adoption and operational reality
Gartner published two predictions in January 2026 that are worth keeping on the table. The first: by the end of 2026, 40% of enterprise applications will include task-specific AI agents, up from less than 5% in 2025. Second, by 2028, 60% of brands will use agentic AI to deliver simplified one-to-one interactions with their customers.
Today's operational picture is somewhat less triumphalist. Gartner's CIO and Technology Survey 2026 indicates that only 17% of organisations have deployed AI agents in production, although more than 60% expect to do so in the next two years - the most aggressive adoption curve among all emerging technologies tracked by the consultancy.
In terms of investment, global spending on agentic AI is forecast to reach $201.9bn by 2026. In parallel, the marketing AI market has reached $47.32bn this year and is projected to reach $107.5bn by 2028, with a compound annual growth rate of 36.6%. And a fact that shouldn't go unnoticed: 88% of marketers say they use AI in their day-to-day work, but only about a third of organisations have scaled it beyond isolated experiments.
Big asterisk: 40% projects could be cancelled
It's not all light. Gartner forecasts that more than 40% of AI agentic projects will be cancelled before the end of 2027 due to unclear value, rising costs and weak governance. This figure explains why, in serious agencies, we don't recommend launching an agent “just because”. What differentiates a project that survives from one that shuts down is rarely the model: it is usually the use case chosen and the discipline with which return is measured.
In Vandelay we see three patterns that predict failure. The first: defining the agent as “the company's ChatGPT”, without a measurable objective. The second: applying it to dirty or poorly governed data; an agent who decides on inconsistent information amplifies the error. The third: not preparing the team to review, vet and correct. An agent without guardrails is a risk, not an advantage.
Real use cases already in production
In recent months, a number of important launches have been announced that illustrate where the industry is moving. Attentive unveiled new agentic capabilities at its Thread 2026 event that analyse customer signals across channels to assess engagement and intent before deciding on the next message. Madhive launched Maverick AI Agents, a suite focused on local video media planning, integrating agentic intelligence directly into the plan.
For a Spanish SME, the most useful cases today tend to be three: the continuous optimisation of bids and creatives on advertising platforms, email and web personalisation based on proprietary data, and conversational customer service where the agent can consult stock, shipping dates or returns without human intervention. Gareth Cummings, CEO of eDesk, recently summed it up: “By 2026, a significant proportion of customer interactions will be agent-to-agent. Shoppers will use assistants to check availability and brands will respond with their own agents.
How to tackle it without throwing money away
Before choosing a tool, three questions should be answered in writing.
FirstWhat repetitive decision, today taken by one person, would it be reasonable to delegate? If you cannot name it in one sentence, you are not ready for an agent.
SecondAre your data in good shape? An agent acting on a CRM with duplicates, inconsistent properties and out-of-date consents will generate more problems than savings. Here the initial investment is usually more in data order than in software licences.
ThirdWhat specific business metrics improve if this works and how much are you willing to lose if it fails? An agent who saves four hours a week but breaks brand consistency does not pass the test.
Once these three questions are answered, it makes sense to start with a single use case, with a narrow perimeter, clear metrics, and a human “in the loop” who approves critical actions for the first few weeks. As the agent demonstrates stability, you can reduce that oversight and expand the scope.
What won't change (even if it seems otherwise)
The strategic part of marketing - understanding the customer, defining the value proposition, deciding where you want to play - is still human work. Agents execute better, measure better and adjust faster, but they don't choose for you who you want to sell to and what your brand stands for. That boundary is what makes having a specialised partner still make sense in 2026: the tool changes, the criteria is still built on data, history and real conversations with your customers.
If you run your own agency, marketing department or online shop and want a concrete roadmap for incorporating AI agents into your operation without falling into the 40% of cancelled projects, write to us at vandelay@vandelay.es or schedule a call from our Artificial Intelligence Services. We audit the use case, the state of your data and the expected return before proposing any deployment - because, as with data, in agents what is not measured well should not be touched.
English (UK) 
Español