Agentic AI – artificial intelligence that can make decisions and take actions with minimal human input – has quickly become one of the most talked-about innovations in B2B marketing.

Promising to handle everything from campaign orchestration to lead qualification autonomously, it represents a step beyond traditional automation or generative AI. The idea is compelling: intelligent systems that do more than respond to instructions and actively pursue goals, adapt to data in real time, and manage tasks across platforms.

In theory, this could be a game changer for B2B marketers grappling with long sales cycles, complex buying committees, and a growing pressure to deliver results with leaner teams.

Those in favour of agentic AI argue it can enhance account-based marketing, personalise messaging at scale, and forecast pipeline performance more accurately than human teams alone.

Vendors have also been quick to position it as the next evolution in marketing technology; and as something that will naturally follow on from the predictive capabilities of LLMs like ChatGPT.

If it’s the job of anyone, it’s the job of marketers to kind of stay on top of trends.

Monica Guan, Rekki

But as with many AI breakthroughs, the reality is proving more complicated. While the ambition behind agentic AI is clear, its practical application is currently patchy.

Questions are already emerging about how autonomous these systems truly are, how well they handle nuance, and whether they will lighten the load for marketing teams, or instead create new layers of complexity.

As hype continues to outpace deployment, marketers are left navigating a familiar tension: bold promises from vendors, limited proof in the business cases being presented that this technology can work at the scale necessary.

Indeed, Gartner anticipates 40% of agentic AI projects will be cancelled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls.

‘Profound implications for the industry’: Mastercard’s CMO on marketing’s agentic AI future

Yet businesses continue to invest millions in testing the water. Nearly a fifth (19%) of respondents to a Gartner poll said their organisation had made significant investments in agentic AI, with a further 42% having made conservative investments. A figure that is only set to rise.

A separate study of more than 1,200 marketers by Outcomes Rocket reveals 48% believe agentic AI will have the “most significant impact” on marketing over the next two to three years.

This poses a risk for Jon White, head of group marketing for fluid power distribution firm Flowtech, as he believes an overreliance on and potential failure of agentic AI “could distract from simply developing brilliant marketing within a B2B environment”.

He says the key thing is not to be blinded by its allure and focus on developing good marketing strategies in the B2B space, regardless of the tools used.

“Every time I go to a conference, every conversation is about AI. And then you have to ask, ‘Yeah, but do you know what you do with it?” he adds.

The need for a clear business case is one of White’s primary concerns with agentic AI at this point in the hype cycle. In his view, even when there is a business case for adopting the technology, most of the justification focuses on efficiency and cost cutting, rather than adding any value.

“[Marketers] are paid to grow and generate additional value, not to trim it down. We’re driven by effectiveness, not efficiency,” he says.

As long as marketers keep this in mind, he doesn’t think agentic will “derail” B2B marketers or cause “any real damage” long term, though, he says.

The question of maturity

Elliot King, vice-president of marketing for revenue content platform Turtl, says he is “surprised” the predicted failure rate of agentic AI projects isn’t higher, considering the maturity of the tech and how many brands are simply “rebranding their existing automations”.

Indeed, the Gartner study finds many vendors of agentic tools are participating in a practice dubbed ‘agent-washing’, wherein they brand existing products, such as AI assistants, robotic process automation (RPA) and chatbots, as agentic despite them lacking substantial agentic capabilities.

“It’s the same with any type of rising trend in the industry […] we saw this in the early 2000s when more businesses started applying simple machine learning,” he says.

He goes on to explain how when a new trend in the sector emerges, vendors will jump on it before their technology meets those standards, leading to an inflection point where a large portion of the technology will fail.

“So I think it’s important that whenever you’re engaging with the vendor that you get the proof,” he says. “You ask them, what are the case studies? What is the ROI like? Can you show me how this works in practice?”

B2B buyers consider ‘less than three suppliers’ in gen AI purchase decision

Peter Bell, vice-president of EMEA marketing at software firm Twilio, says the potential for overpromising and underdelivering technology should be somewhat expected at this stage of the game.

“We’re definitely in the hype cycle. I think there’s a real fear at senior levels that this could mean you go out of business […] and people do jump on the bandwagon rightly or wrongly,” he explains.

Bell believes the companies adopting agentic AI that will fail will be the ones that are “overly ambitious and very broad in scope”. For example, replacing customer marketing with agentic AI, “that is something that will probably fail”, he says.

“However, if you want to use agentic AI to do something like deal with all password resets, that is more likely to succeed because it has a reasonable scope and a clear goal.”

Despite all this, he remains optimistic about the future of the technology and where it may go.

“It’s something where I see huge promise. It’s just a question of we’ve got to crawl, then walk, then run.”

Responsible use and ethics

As is the case with any emergent technology, concerns have been raised about how to use agentic AI safely and what guardrails need to be put in place around data protection.

Monica Guan, head of marketing for restaurant supply management firm Rekki, believes that as marketing is the “face” of a brand, it has more responsibility than other teams to use agentic AI ethically.

“We’re what the public sees about the company,” she says. “[Agentic AI] affects your brand directly versus other teams, such as sales, which might just use it as a tool.”

However, in her own experience of buying from AI vendors, very few have a large focus on ethical use, she says.

“If it’s the job of anyone, it’s the job of marketers to stay on top of trends… I think it’s a space [agentic AI] to proceed with caution, but something we have to stay on the pulse of.”

We’re driven by effectiveness, not efficiency.

Jon White, Flowtech

To that end, Nicole Denman Greene, vice-president and analyst for artificial intelligence, content, and digital experience at Gartner, says the most important thing for marketing leaders right now is to get their people ready to adopt agentic AI correctly and safely.

She says: “Companies now need to think about responsible use, data, hygiene, security, and guardrails like governance.”

Education and adopting the technology in a measured way will be key to a successful rollout of agentic AI more broadly in the future, she adds.

“There’s no magic AI pixie dust. It’s not going to solve all of the problems. You need the discipline, you need the strategy, you need the training and upskilling to be successful using this tech.”