Agentic AI

Artificial intelligence is already changing how digital advertising is planned, bought and measured. The next development is now coming into focus: Agentic AI.

While traditional AI primarily supports people by analyzing data or generating recommendations, agentic AI can take the next step by acting on those insights and carrying out tasks within defined parameters. For advertising, this could fundamentally change how campaigns are planned, activated and optimized.

But agentic AI should not be seen as a technology that will suddenly replace today's programmatic ecosystem. Instead, it is likely to become an additional layer that connects data, platforms and decisions more efficiently.

What is agentic AI?

Agentic AI describes AI systems that can work towards a defined goal with a certain degree of autonomy. Instead of simply responding to individual prompts, an AI agent can interpret a task, determine the necessary steps, use connected tools or data sources and take action based on predefined rules.

For example, a traditional AI assistant might analyze a campaign brief and recommend audiences, channels and budgets. An agentic system could go further by turning the brief into a media plan, identifying suitable inventory, setting up parts of the campaign and monitoring its performance.

This does not mean giving AI unlimited control. In professional environments, agents can operate within clearly defined permissions, budgets, thresholds and approval processes.

How can agentic AI be used in advertising?

The advertising industry is particularly suited to agentic AI because programmatic advertising already relies on automation, data and clearly defined processes.

Agentic capabilities are emerging across different stages of the media lifecycle. These include:

• Interpreting campaign briefs and generating media plans

• Discovering audiences, inventory and suitable deals

• Configuring and trafficking campaigns

• Moving budgets within predefined thresholds

• Monitoring delivery and diagnosing problems

• Consolidating reporting across platforms

• Producing optimisation recommendations

• Adapting creative assets based on performance and context

Agents can also identify optimization opportunities and recommend changes based on campaign performance. Over time, they may increasingly be able to implement those changes themselves, provided they remain within predefined parameters.

Some of these capabilities are already available within individual platforms or controlled ecosystems. There have also been early examples of campaigns being planned and activated through agentic interfaces, including in Connected TV.

More about CTV

Three stages of agentic advertising

The development of agentic AI in advertising can broadly be divided into three levels of maturity.

1. AI assisted planning and insights

This is already happening today. Conversational AI can interpret campaign briefs into recommendations, create reports and reduce manual work.

For media teams, this means less manual work and faster access to insights. Instead of navigating multiple dashboards or manually consolidating data, users can increasingly interact with platforms through natural language.

2. Agentic campaign operations

Agents can increasingly perform actions such as campaign setup, deal discovery, budget changes and reporting, provided they operate within explicit permissions, thresholds and human approval processes.

Agents can increasingly perform actions such as campaign setup, deal discovery, budget changes and reporting, provided they operate within explicit permissions, thresholds and human approval processes.

3. Fully Interoperable agent-to-agent buying

The longer term vision is a system in which a buyer's agent can communicate directly with platforms and other agents across the advertising ecosystem.

In this scenario, a buyer agent could understand a campaign objective, identify relevant audiences and inventory across multiple platforms, compare opportunities and activate campaigns across DSPs, SSPs, publishers and data providers.

For a buyer agent to communicate reliably with multiple DSPs, SSPs, publishers and data providers, the industry still needs common schemas, identity and authentication, governance, auditability and widespread platform support.

Will agents replace programmatic advertising?

Probably not. The real time auction itself is unlikely to disappear. Programmatic systems are built to make extremely fast, deterministic decisions at impression level, and there is little reason for agents to replace that infrastructure.

Instead, agents are more likely to operate around the auction. They can determine strategy, interpret objectives, select audiences and inventory, manage budgets and react to performance, while existing programmatic systems continue to execute transactions at scale and speed.

In this sense, agentic AI is less about replacing programmatic advertising and more about changing how humans and machines interact with it.

What comes next?

The future of Agentic AI

In the short term, AI assistants will increasingly become part of everyday media planning, analysis and reporting. From there, agents will manage larger parts of campaign and deal workflows within clearly defined permissions.

Over the longer term, we can expect more autonomous cross channel planning and buying. But even in a highly automated ecosystem, human strategy will remain important. Brand objectives, budgets, compliance requirements and business decisions still need to be defined by people.

The future of agentic advertising is therefore unlikely to be about handing complete control to AI. It is about creating systems that can take on increasingly complex tasks while remaining transparent, controllable and aligned with the objectives they were given.

More about advertising with AI