Marketing operations are increasingly conducted through AI-mediated systems, according to McKinsey & Company. Agentic AI is beginning to shape marketing workflows as consumers use digital platforms to discover and purchase products, while expectations for personalisation and response times continue to tighten.
Generative AI tools are already being used for tasks like copywriting and image creation. These deployments are often limited to isolated use cases, resulting in fragmented systems that increase output volume without improving overall business performance.
McKinsey describes this as a gap between widespread experimentation and limited enterprise impact, driven in part by disconnected pilots that do not integrate in workflows. Existing marketing technology stacks – including content management systems, digital asset management platforms, customer relationship systems, and analytics tools – were not designed for shared data models or real-time agentic workflows.
Agentic AI and workflow redesign
Agentic AI systems capable of executing multi-step processes are built on foundation models. Systems allow organisations to structure workflows where AI agents handle execution while human teams supervise outcomes. In this model, a single marketing professional can oversee multiple agents responsible for tasks like content generation and optimisation. The report describes this structure as a hybrid human – agent workforce, where humans define objectives and guardrails while agents carry out execution in multiple steps.
The report states that adopting this approach requires unified data layers, consistent identity frameworks, and systems that allow agents to interact through application programming interfaces. The report notes that system interoperability, not model ability, is often the primary constraint in deploying agentic workflows. It adds that flexible model-serving infrastructure and activation systems capable of exposing reliable APIs are required to let agents act in content and distribution environments.
McKinsey estimates that agentic AI could support up to two-thirds of current marketing activities, including synthetic audience testing, where AI-generated audience simulations are used to evaluate campaign performance before deployment, with automated content generation and audience-based media planning. The firm also reports that organisations implementing these workflows have recorded potential revenue increases of 10 to 30% through more targeted execution.
Agentic systems can also accelerate campaign processes by a factor of 10 to 15, including idea generation and deployment. The report states that automation of operational tasks allows marketing budgets to be reallocated from internal processes toward direct customer engagement.
Implementation remains limited. Data cited by McKinsey indicates that nearly 90% of chief marketing officers are testing AI applications, while fewer than 10% have deployed end-to-end workflows that generate measurable value. The report attributes this gap to the complexity of redesigning workflows and integrating systems, not limitations in the underlying AI models.
Designing agentic marketing workflows
Organisations adopting agentic AI are restructuring workflows by mapping existing processes into detailed task structures. This includes identifying dependencies on systems like CRM platforms, digital asset management tools, and analytics pipelines. Some companies have broken workflows into hundreds of micro-tasks to identify where automation can be applied. The report notes that this mapping also includes insight-related activities like data synthesis, hypothesis generation, and interpretation of consumer signals, which remain partially dependent on human judgement.
Tasks are then grouped into functional categories like data analysis, content generation and execution. In one example cited in the report, a consumer brand classified marketing activities into reusable agent archetypes.
The report states that this organisation identified nearly 100 modular agents in content-related workflows. These archetypes included functions like content generation, knowledge retrieval, localisation, analysis and execution, letting agents be reused in different marketing processes.
Implementation also depends on system compatibility. Integration challenges often arise when connecting agents to data platforms and content repositories. Some vendors, including Adobe and HubSpot, have introduced embedded AI agents in marketing platforms to generate and update content based on real-time inputs. These agents can tailor content variations, update assets in channels, and respond to behavioural signals without requiring manual intervention at each step.
Workflow redesign changes the role of marketing teams. Responsibilities include validating outputs, managing data quality, and maintaining compliance with brand and regulatory requirements. Teams are also responsible for overseeing content metadata, orchestration rules, and API governance to ensure agents operate consistently and safely. Human roles also include reviewing AI-generated concepts, refining outputs, and ensuring alignment with brand positioning and market context.
Organisations are investing in abilities like prompt engineering, quality monitoring, and data and AI fluency to support these workflows. These functions help manage agent performance and ensure outputs align with business objectives. Additional abilities include applied machine learning, experimentation design, and workflow orchestration to support continuous optimisation.
Deployment is typically phased. One consumer brand implemented its agentic marketing system in three stages: an initial phase focused on continuous ideation, a second phase introducing automated pretesting and brand and risk checks, and a third phase extending localisation and market rollout. The report states that this phased approach allows organisations to prioritise high-impact workflows while preparing underlying systems for broader deployment.
Early pilot results show reductions in production timelines. In some cases, content creation cycles were completed up to four times faster than traditional processes. Agentic systems are also being applied in media execution, where AI agents adjust campaign parameters like budgets and creative variations in real time. These systems can perform continuous optimisation by making incremental adjustments in campaigns, reducing the need for manual intervention.
Governance and implementation challenges
Governance remains a important consideration due to the direct impact of marketing on consumer-facing content. Survey data cited by McKinsey identifies brand and legal governance, ability gaps, technology under-investment, and data bottlenecks as primary concerns among marketing executives. The report also highlights the need for validation mechanisms to ensure AI-generated insights meet defined accuracy thresholds before being used in decision-making.
Agentic AI is being deployed with other automation technologies, including robotic process automation and machine learning systems. The report notes that organisations are evaluating these tools collectively not relying on agentic systems alone. It adds that focusing exclusively on agentic AI may limit efficiency gains if other automation approaches are not integrated into workflows.
Current implementations combine automated execution with human oversight to manage operational complexity and maintain control over brand and compliance requirements.
(Photo by Lukas Blazek)
See also: The world’s largest ad holding company just bet its AI future on one vendor
Find out more about the Digital Marketing World Forum series and register here.
