AI agents work smarter when they work together

Meta plans to have AI agents build creative, target audiences and optimize campaigns to create a fully automatic campaign workflow. Amazon is implementing a modular approach to agentic AI, leveraging it for creative, targeting and optimization as well for similar purposes.

It’s no secret that walled gardens are betting on AI to make digital advertising more efficient — but the question is, for whom?

The benefits of agentic AI within walled gardens will largely be for the walled gardens themselves, as they reduce internal costs and attract and retain long-tail advertisers with limited resources.

Enterprise advertisers, however, will soon find that agentic AI within walled gardens is opaque and difficult to integrate and compare with the rest of their marketing efforts. For these marketers, the real benefits of AI agents will come from outside the walled gardens. Marketers who want AI to work on behalf of their goals need to deploy agents that they can control, rather than relying on agents whose primary mission is to help a walled garden. 

Silos on overdrive

Consider a typical marketing goal to achieve an incremental increase in sales. Today, marketers piece together insights from a variety of campaign partners, including walled gardens. They pull what campaign insights they can and compare outcomes across their entire media buy. This is imperfect; marketers lack full transparency and don’t have ownership over audience data on partner platforms.

But if independent AI agents within walled gardens start making decisions on a marketer’s behalf, marketers will lose additional visibility and control, making cross-portfolio optimization far more difficult.

Meanwhile, dominant tech and media companies will fight for more campaign budget, but they cannot optimize campaign goals holistically.

In some cases, tech giants will embrace collaboration early because it benefits them. For example, they will want to ingest campaign goals, creative guidelines and target audience personas because it will tether marketers to their platform and help them improve outcomes. In other cases, tech giants will resist true collaboration — even as they invite marketers to share data with them. They will neglect to share data back, weakening marketer agents and strengthening walled garden agents. 

This is our industry’s current scenario on overdrive, where marketers are frustrated and media companies are undercut by middleware. 

The way forward: Independent, insightful, interoperable

Independent AI agents (i.e. agents that aren’t attached to a specific platform or walled garden) can work across marketers’ entire advertising strategies to gain significantly more insight and efficiency. When trained on relevant use cases and data, agents can act on behalf of a marketer, using their unique goals to analyze, organize, measure and drive outcomes across a media plan. Companies can also use agents that are related to, but outside of advertising and marketing — including agents to manage inventory, improve customer service, drive sales, improve their supply chain and much more.

As marketers build out their own fleet of agents, tech and media companies will deploy agents of their own to improve yield optimization, data sharing and workflows — to name a few.

In our highly integrated industry, the future will be significantly more productive if we build these agents collaboratively. In a scenario where marketers’ agents can talk to partner agents and share desired outcomes, ideal audience targets and performance information, tech and media companies would be judged on their true contribution and could start building solutions and creating content that drives even higher performance.

Of course, marketers, their agencies and publishers all have their own interests in mind — and sometimes have opposing goals (media companies will always want a higher CPM price than advertisers want to pay). But this is where AI agents will be at their best, running negotiations with a wealth of knowledge on performance and trade-offs, rather than with a myopic view of a particular platform.

Writing the future

The advertising industry must act swiftly to establish agent-to-agent communication standards before fragmented, proprietary systems become entrenched.

Embracing open standards and turning experimentation into scalable, enterprise-grade platforms will be key. Open protocols like Model Context Protocol (MCP) facilitate seamless collaboration between marketing agents and platform agents. By enabling intelligent, highly specialized agents to interoperate across clouds, frameworks and media ecosystems, the industry can move faster, work smarter and build more connected, future-ready solutions. 

With agents designed to leverage open protocols like MCP, for example, a brand marketer could deploy an AI agent that seamlessly coordinates campaign execution across media buying platforms, creative testing tools and retail data partners without being locked into a single vendor ecosystem. That kind of interoperability would enable real-time optimization, faster go-to-market and smarter budget allocation — all powered by collaborative agent infrastructure.

But speed is critical here. The AI landscape is evolving at a breakneck pace, and the companies that establish interoperable agent ecosystems first will gain a significant competitive advantage. Those who delay risk being locked into closed systems that ultimately undermine campaign performance and limit strategic flexibility.

The window to shape this future is narrow, but the opportunity to transform advertising through truly collaborative AI is unprecedented. Companies should begin pilot programs with MCP and A2A protocols immediately, working together to establish the technical and ethical frameworks that will define agent collaboration for years to come. If we don’t implement the right collaboration framework, there is a real risk that companies use agentic AI to build higher walls around their gardens, while marketers and media companies lose ground. 

To be clear: Agentic AI is happening. Whether agents work collaboratively is up to us. Right now, we have an opportunity to write our own future. 

This op-ed represents the views and opinions of the author and not of The Current, a division of The Trade Desk, or The Trade Desk. The appearance of the op-ed on The Current does not constitute an endorsement by The Current or The Trade Desk.