"AI Agents in marketing: opportunities and risks"

Are AI Agents, like OpenAI Operator, a revolution in marketing?

Although I’m an AI enthusiast myself, I try to remain cautious about getting overly excited about new technologies. Some LinkedIn experts have already written "this changes everything." I, on the other hand, will say that we are entering very interesting times, but the impact of Agentic AI itself depends on skillful implementation. And that responsibility lies with both solution providers and marketers themselves.

Maybe before we get into the details, it’s worth emphasizing that this shift has two vectors: consumer and business.

In the first case, we’re already talking about AI agents that can autonomously perform specific tasks on behalf of users - such as booking a hotel. That’s one of the use cases for OpenAI’s Operator, which was showcased by Booking itself. Popularization in this area could, in the long run, lead to the erosion of the traditionally perceived consumer journey. Because if a rational robot is making purchases on our behalf, does it care about emotions in communication? Or only about price and the quality of the offer?

In the second case, we’re talking about enterprise-level solutions, where marketers can use autonomous tools for their own needs. For example, commissioning competitive research, planning and executing a media campaign.

How exactly can OpenAI’s new development improve the work of marketers and agencies?

This ties into the second area. I have the privilege of being the creator of one of the first Agentic AI applications for marketing in Poland - Adaily 2.0. What we’ve built is a complex system that, based only on "brand" and "country," can generate full consumer, category, competition and trend research and then, based on that, propose a complete strategy with an action plan. In under a minute.

That’s why the benefits are obvious.

First, we’re talking about a clear time saving. AI agents can identify hundreds of online sources faster and more accurately than humans.

Second - exploring new directions. As humans, we are often biased in how we view marketing. We may lose perspective on what we’re doing (for example, because we’ve worked with the same brand for 10 years) or have our own preferences. AI agents, on the other hand, can evaluate thousands of alternatives at once and recommend the most effective paths.

Third, it’s a redefinition of the employee’s role. For some time now, people have been saying that when working with AI, we become curators rather than creators. With Agentic AI, this will become even more evident: soon, what will matter more is how we assign tasks to the agent and how we evaluate its work, rather than how we complete the work ourselves.

Who will benefit the most from their introduction?

In my opinion - clients. But only in the long run.

Right now, the main beneficiaries of Agentic AI are advertising agencies. Using these solutions allows them to increase their output while maintaining quality. They can participate in 10 pitches simultaneously and have well-prepared proposals for each. With the same team as before. The savvier agency owners will be able to redefine the business model in a way that maximizes the return on the agency’s biggest asset: creativity, by combining human and machine efforts.

However, as Agentic AI technology develops and AI adoption increases on the client side, that’s where we’ll see the biggest marginal benefits. Marketers operate within the 4Ps framework and their work doesn’t start or end with communication. They have to divide their attention. They rely on agencies for advertising, which provide expert know-how and the ability to execute complex initiatives.

But what happens when Agentic AI gradually, yet effectively, begins to take over these areas?

At the end, an interesting fact. Two weeks ago, when I posted on LinkedIn about our Agentic AI solution for marketing, I received messages from over 600 people asking for details. Most were agencies, but there was also a significant number of clients. And the first company I demoed our system to was a large FMCG company operating in multiple segments across several markets.

Is this the beginning of a new era? We’ll see.

What risks and challenges come with using AI Agents?

I think the biggest problem with Agentic AI is the same as with AI in general: a lack of adaptation to a specific industry.

Can a regular AI system identify a strong consumer insight? Probably not. But in the hands of an expert or as a well-designed tool created by industry practitioners - yes, it can. That’s exactly how we built our system: tailoring it to the specific needs of marketing professionals.

Of course, the same professionals could be power users and build their own systems that complete the right tasks, in the right way, in the right sequence. The question is - who has the time for that alongside their daily responsibilities?

However, if these systems are good enough, adoption will be exponential. No one will be able to ignore the potential business benefits.

To end with a more philosophical question.

What if consumers start massively using AI agents for their purchasing decisions?

What clothes to choose, where to order food delivery, or where to book a vacation.

Personal AI agents will query business AI agents for conditions and maximize benefits for the consumer within given parameters.

No one will watch ads. No one will click on banners.

A few weeks ago, Aravind Srinivas, founder of Perplexity, one of the leading AI-powered search tools, outlined this vision. Today, Perplexity already has its own AI agent running on Android phones.

Dystopia? Utopia? I’ll let everyone decide for themselves.

A short interview I did for Polish media outlet NowyMarketing. Original article in Polish here. The still comes from Metropolis (1927) - watch it if you have a chance!