The exponential growth of generative AI tools like ChatGPT and Gemini have quickly become mainstream in digital marketing. Taking the next step leads users into a world of Agentic AI, where systems can autonomously execute multi-step, goal-driven workflows.
However, this will require careful planning, upskilling and some very secure guardrails. But it’s worth it.
Still early days
GenAI tools have been widely adopted by marketing teams, and profit-focussed leaders should now be eager to embrace the promise of Agentic AI. However, at a recent gathering of 250 digital marketing professionals, a quick poll of the audience showed that while nearly all used some form of AI, only two had ventured into Agentic AI workflows.
This surprisingly low uptake reveals the many real (and some imagined) barriers that must be addressed before Agentic AI can become as ubiquitous as other marketing technologies.
There are a number of reasons for this limited uptake. Building and implementing Agentic AI solutions is different from deploying off-the-shelf AI tools. With a chat LLM (large language model), you sign up, ask a question, and get an answer back with no real technical background required.
Agentic AI, however, requires you to design and integrate custom workflows that can reason, act, and learn according to your specific business goals.
This complexity is further exacerbated by limited awareness. Many marketing leaders remain unsure of what Agentic AI really is, let alone how it might help their business. Concerns about cost, risk, and resourcing also quickly emerge when speaking to industry professionals, especially when compared to the apparent simplicity and affordability of existing cloud-based AI platforms.
Unlocking the practical potential
Despite the obstacles, the potential of Agentic AI to transform digital marketing should not be underestimated. At our company, we’ve begun rolling out Agentic AI workflows that optimise product feeds, ensuring that titles and descriptions better match consumer search intent.
In one case study we saw a 36% increase in CTR with a 44% decrease in CPC.
This isn’t magic, it’s the result of autonomous agents monitoring real-time data, analysing performance, and tweaking content to maximise relevance and click-through.
Beyond product feeds, Agentic AI is showing promise in campaign planning and ongoing optimisation. Imagine a system that not only monitors market trends and consumer behaviour, but dynamically adjusts targeting, creative content, and bidding strategies in response to new data as campaigns unfold. The early results we’ve seen, while still building toward larger datasets, point toward substantial efficiency gains.
Challenges on the road to Agentic AI
The path to adoption won’t necessarily be straightforward. Agentic AI requires expertise in workflow design, integration, and oversight, and skills are currently in short supply outside specialist agencies or advanced tech teams.
Ethical and governance challenges should also not be overlooked. To function well, Agentic AI must often ingest and process significant volumes of personal or behavioural data, raising questions about consent and ownership.
Besides ethical challenges, we must also consider security. Key questions include: Who can access the agent? And what data can the agent access? These questions are very important, but there are solutions. For instance, an agent can use the access levels of the current user, ensuring that users only access data they are authorised to see. This approach also allows multiple users to safely use the same agent.
It’s essential to install strong guardrails or mechanisms that ensure the AI operates within well-defined ethical boundaries. Marketing leaders should also ensure the system functions as transparently as possible. This includes showing clients how safeguards are implemented and how the system responds when outputs move outside approved areas.
Then there’s the regulatory angle. Much like the introduction of GDPR sharpened the industry’s focus on data protection, the coming wave of AI regulation (from the EU’s AI Act to evolving national guidelines), will force companies to have a clear strategy for the technology.
A smarter, more democratised future
Looking ahead, it’s clear that like all AI technologies, Agentic AI will get smarter and more powerful. But we believe it will also be more democratised, becoming available to users far beyond technical specialists. The workflows will be less rigid, allowing marketing teams to tackle broader, more open-ended challenges.
For now, the first-mover advantage will only be available to the bold, but the likely future will be one where Agentic AI becomes a standard tool across industries, rather than the preserve of a tech-savvy few.
In addition, the human factor will (and must) always matter. AI may handle the repetitive and complex, but the spark of creativity, ethical judgment, and strategic oversight must come from people. As Agentic AI tools become ubiquitous, marketing leaders will need to think harder than ever about the skills, processes, and values that determine their use.
It’s not a cure-all
It’s important not to implement AI just because it’s the latest shiny tool. Rather, identify where you need a step change in efficiency, insight, or performance, and design your Agentic AI solution to tackle that problem specifically.
Partnering with agencies or experts with proven experience will undoubtedly help accelerate your journey and prevent costly missteps.
Used correctly, Agentic AI has the potential to drive efficiencies into your business, quickly becoming a force multiplier. But this can only happen if it is embraced thoughtfully and responsibly.
Quintijn van Kessel is head of product and innovation at Incubeta.
