From prompts to action: what agentic AI could mean for agencies and marketers
We’ve become familiar with asking AI to research a topic, summarise information or suggest a recommendation. But what happens when it can also act on that recommendation?
Last month, I attended AgenticAdvertising.org’s Sydney session on 22 October, where the conversation explored this next stage of AI in media: agents that can work towards an objective, interact with tools and platforms, and take action within agreed boundaries.
For agencies and marketers, the potential is significant. Much of our working day still involves gathering information from disconnected systems, coordinating between platforms and moving processes forward manually. Agentic AI could change how that work gets done, and where we spend our time.
Moving from individual tasks to connected workflows
The distinction is between helping with a task and progressing a workflow.
An AI tool might help a trader analyse campaign performance and identify an optimisation opportunity. An agent could potentially assess the available options, recommend a course of action and execute the next steps, within approved parameters and with human oversight where required.
That could mean less time moving information between systems and more time evaluating what it means for the client.
One development worth watching is communication between agents. AgenticAdvertising.org’s Ad Context Protocol, or AdCP, provides a shared language for agents to work across advertising platforms, supporting activities such as discovering inventory, buying media and measuring delivery.
The promise extends beyond automating isolated tasks. It could connect parts of the media process that currently depend on emails, spreadsheets, manual comparisons and separate platforms.
What could this change for agencies and clients?
For agencies, the clearest opportunity is reducing the operational work surrounding good media decisions.
In planning, agents could help gather audience research, market insights and relevant media opportunities against a brief. In trading and activation, they could support inventory discovery, proposal comparisons, campaign setup and optimisation. Across campaign delivery, there is potential to reduce repetitive work in reporting, trafficking and quality assurance.
These are areas where teams can spend substantial time assembling the information they need before they can apply their expertise.
For clients, that could translate into faster turnaround from brief to activation, clearer comparisons between media opportunities and more responsive optimisation. It could also create more space for conversations about business performance, rather than the mechanics of getting a campaign live.
Those benefits will depend on the quality of the information agents receive, how well platforms connect and the controls around their decisions. Faster execution only helps if the work is moving towards the right objective.
Where does that leave people?
One recurring question during the session was whether agentic AI will replace roles across agencies and marketing teams.
My takeaway was that, as more operational tasks become automated, the skills behind good decision-making become even more important.
An Optus panellist reinforced the importance of the people directing the technology: their industry knowledge, business context and strategic understanding.
For traders, planners and strategists, that means asking better questions, interpreting outputs, challenging recommendations and recognising when an answer misses something important.
An agent may be able to compare proposals against a set of criteria. Deciding whether those criteria reflect the client’s actual challenge still requires judgement. So does understanding the nuance behind a brief, navigating commercial relationships and taking responsibility for a recommendation.
Brand safety, accountability and client trust remain central, however capable the technology becomes.
Giving better thinking more room
What stayed with me was the possibility of changing how we use our time.
If agentic AI can reduce the effort spent gathering, transferring and checking information, agencies have an opportunity to put more attention into interpreting it, developing stronger recommendations and working more closely with clients.
The question for marketers is where that would make the greatest difference: which processes would benefit from automation, which decisions need human approval, and what better use could we make of the time saved?
The real value will come from turning that capacity into better decisions and stronger outcomes for clients.