The Model Context Protocol (MCP) makes it possible to exploit data at an unprecedented level of granularity, much more than it seeks to unify or simplify the advertising market.
Let’s be clear: a new attempt to unify the media market is not for tomorrow, and the MCP is already proof of this. Several protocols emerge in parallel, universality is called into question from the start, and each actor develops its own approach, its environment, its own logic. As Paul Ripart, deputy digital and data director at Prisma Media, points out, the risk is even to recreate walled gardens, but at a more invisible level.
Thinking that we will be able to manage a multimedia campaign in a completely unified way is not realistic. Purchasing logics remain too different. Technical constraints, formats, rules specific to each platform prevent any complete centralization. This is not where the transformation takes place.
The real potential: micro-tasks for micro-data
The real potential of MCP is micro. It is in the execution of micro-tasks, on a large scale: categorizing actions, applying a coherent and enriched taxonomy, correctly generating UTMs, structuring campaigns according to naming conventions and predetermined fields, placing the right information in each of the platforms.
These are simple, but repetitive and massive tasks. And today, they still rely largely on humans. MCP makes it possible to automate them, make them more reliable, and execute them in a consistent manner. It is concretely the generation of structured data, micro-task by micro-task.
Unknit to find granularity
The problem today is that we have an extremely rich media fabric… but completely tangled. All the data is there. All threads exist. But they are intertwined, superimposed, difficult to read. To find value, you have to agree to unravel. Return to the thread. Understand every campaign, every piece of data, every interaction. Find the granularity.
It is only at this level that we can really exploit the data. And this is precisely where MCP comes in handy. Not to unify, but to activate these threads one by one, at the right scale. Once this work is done, we can weave again. Recreate a coherent, readable, activatable mesh.
Moving from API logic to usage logic
Another key point is the ability to move beyond APIs. There is a lot of focus on MCP, and it makes sense with its strong capacity to interconnect the media ecosystem. But agentics goes beyond MCP. The duo Claude Cowork and Claude for Chrome, which combines a Browsing Agent which takes control of your Chrome, and Claude Cowork which directly manipulates the APIs to which it has access, is a total upheaval of the media ecosystem. Today, many workflows rely on complex technical integrations, often limited, and which each live in their SaaS, each in their browser tab. Agentics in its entire range, from MCP to virtual action, allows direct and universal interaction with all interfaces.
This enrichment of the agentic range takes us from a technical logic to a logic of use. This complete agentic range completely changes the way campaigns can be run.
An essential discipline on data
But this mechanism only works if the data is clean. The MCP does not correct anything. He executes. Without strong data hygiene, it will simply reproduce and amplify errors. We must therefore impose very clear discipline, which extends to very concrete rules: account presets, prohibitions, purchasing frameworks. A form of toll which imposes rigor in execution. Without this, automation is pointless.
Concrete, but targeted interoperability
The interest of MCP is in interoperability, but not the one we imagine. It is not a question of creating a single system, but of allowing agents to coordinate categorized actions between tools which live in a plurality of platforms.
This could, for example, consist of connecting media planning with tools like Asana or Monday, automatically generating a media plan from a brief, assisting the briefing or distributing data across platforms. Interoperability which allows you to work in the interface of your choice; : create a media plan from Claude within a media operations platform, or from this same platform by taking advantage of Claude’s agentic strength. The main thing: respect the practices.
Better exploit complexity, not eliminate it
The media market is not going to get any simpler. But it can be better exploited. MCP and agentic AI more broadly with agentic browsing, makes it possible to decomplexify operations, organize actions, optimize each purchase by task and make data activatable.
On condition of accepting a simple reality: we will not resolve complexity through unification. We will make it usable by the structure. The subject is not to connect more and more, but to better organize what already exists. To avoid automating chaos.
