If you thought programmatic advertising was the final frontier of media buying innovation, think again. The next big shake-up is coming at you with “agentic AI” – essentially AI “agents” that can take on tasks autonomously.
Two new tech protocols are leading this charge: Anthropic’s Model Context Protocol (MCP) and Google’s Agent-to-Agent (A2A). This new tech sounds super-nerdy, but their impact on how we plan and buy media could be huge.
In this post, we’ll break down what MCP and A2A actually are (don’t worry, in plain English), how they work compared to today’s process, and what it all means for you as a media professional.
MCP: A Universal Adapter for AI and Data
Model Context Protocol (MCP) is an open standard that lets AI systems plug into your data sources and tools seamlessly. In fact, Anthropic calls it a “USB-C port for AI applications,” meaning it provides a universal way to connect AI models to different databases, content libraries, business apps, and so on.
Instead of building custom integrations for every single platform or dataset, you connect once via MCP and the AI can securely access what it needs.
MCP Consolidates Your Data
In simple terms, MCP gives an AI assistant the context it needs from all your scattered data.
Imagine your media planning AI could, with one adapter, tap into your CRM for audience insights, your Excel sheets for budgets, and your campaign performance dashboards for history – all without special coding for each. That’s what MCP enables.
MCP standardizes how applications feed information to AI models, replacing today’s jumble of one-off integrations with one consistent protocol. The result: your AI tools can finally break out of their data silos and work with a richer picture of the world.
MCP Example
For example, if you asked an AI agent (like Claude) to draft a media plan, MCP could allow it to instantly pull in CRM data, media brief, advertiser profile, past campaign results, and placement options from your systems to inform the plan.
Currently, that kind of data merge is often manual – someone pulls reports or uses a custom API. With MCP, it’s as if the AI has a universal plug to directly access any data source you’ve approved, making its recommendations far more informed and useful.
A2A: A Common Language for AI Agents
While MCP lets AI tap into data, Agent-to-Agent (A2A) lets different AI agents talk to each other.
A2A is an open protocol from Google that serves as a common language for AI agents to communicate, share information, and coordinate actions, even if they were built by different companies or for different tasks.
Think of a team of digital assistants – one could specialize in planning, another in buying, others in research – and A2A is the messaging system that lets them collaborate in real time.
A2A Enables Agents to Work as a Team
In essence, A2A turns isolated AIs into a team. With A2A, an AI agent can find other agents with the skills it needs and “ask” them to help.
For example, your planning agent could call on a social-media-buying agent to execute a portion of the plan, or a research agent to fetch the latest audience trends. They speak a standardized protocol to securely exchange data and requests, almost like giving AI agents a common API to work together.
A2A is Open and Aims to be Secure
Importantly, A2A is designed to be vendor-agnostic and secure. It doesn’t matter if one agent is built on Google’s tech and another on a different framework – if both speak A2A, they can collaborate. Google developed A2A with more than 50 partners, from software giants to consultancies, to ensure it can work across many systems.
In practical terms, A2A could allow, say, your AI marketing assistant to negotiate with an AI-powered ad inventory system or publisher’s agent directly. They could negotiate prices, placements, or share targeting info without a human middleman typing instructions, because the agents themselves handle the dialogue.
How Does A2A differ from APIs?
A2A is more dynamic. A2A doesn’t require pre-defined integrations between each pair of systems. Instead, any A2A-enabled agent can find and talk to any other, on the fly.
It’s like giving each AI its own email address and common language, so they can discover each other and cooperate.
For media buying, that hints at a future where your various platforms’ AIs – search, social, display, etc. – might coordinate with each other automatically through A2A, rather than operating in isolation.
How Is This Different from Today’s Media Buying Process?
Today’s media planning and buying process, even with programmatic tools, still involves a lot of manual and siloed work. You might use one system for planning (spreadsheets, anyone?), another for buying on each channel (DSPs, social ad managers), and yet another for analytics – and they don’t naturally talk to each other. If you’re lucky, you have some APIs or integrations, but those are often custom and brittle. For the most part, humans act as the “glue” – exporting data from one system, uploading to another, interpreting reports and then adjusting plans by hand.
MCP and A2A Work Together to Automate Your Tasks
With MCP, an AI-driven tool can automatically pull in context from all your systems as it works, instead of waiting for you to feed it files. With A2A, the AI running your campaign could directly interact with the AI on the ad platform or the AI analyzing performance, rather than waiting for a person to mediate.
Applications in Campaign Optimization
You (the human) might notice mid-flight that one channel is doing better and manually shift budget, or you rely on a platform’s auto-optimizer (which only optimizes within that platform).
In a future with agentic AI, your buying agent could chat with a creative-generation agent and an analytics agent in real time – e.g., “Creative X isn’t performing on Channel Y, let’s swap in Creative Z that our generative AI just made, and while we’re at it, reallocate $ 5K from Channel Y to Channel Z where our ROI is higher.” This would happen across platforms through A2A, without needing each platform’s UI in front of you.
MCP and A2A Bring Much-Needed Standardization
Another key difference is standardization. Today, every ad tech platform has its own way of doing things.
By contrast, MCP and A2A aim to make connections universal. MCP replaces many bespoke data connectors with one protocol, and A2A sets a single lingua franca for agent communication. These protocols could reduce the heavy lifting in getting systems to work together.
You wouldn’t need to write a custom integration for each new AI tool or ad platform; if they’re MCP/A2A-compliant, they’re plug-and-play in your workflow.
In short, today’s process is like having a bunch of employees who all speak different languages and need a translator (you) to coordinate. The MCP/A2A approach would be like teaching all your tools to speak one language, and even to take initiative to help each other. Less back-and-forth for you, more auto-pilot assistance from AI.
How Could MCP and A2A Transform Media Planning & Buying?
So, what does this look like in action? Here are a few concrete ways MCP and A2A could shake up our day-to-day media planning and buying.
Always-On Research and Planning
Picture having a junior media planner bot who never sleeps.
Using MCP, it can pull in up-to-the-minute consumer data, competitive spend info, and audience insights from all your sources. Using A2A, it could then ask a forecasting agent to run dozens of scenarios. By morning, you have a media plan draft waiting in your inbox – complete with rationale – that the AI team hashed out overnight. You wake up and simply tweak the strategy.
It’s like giving your team a tireless intern who already knows where to find every file in the cabinet.
Cross-Channel Optimization on Autopilot
Today, you might use Google’s Performance Max or Meta’s Advantage+ as isolated tools that optimize within their own ecosystems. These are examples of agentic-like products, but they don’t coordinate with each other – you are still the coordinator.
In the future, you could have an overseer AI agent that uses A2A to coordinate all those channel-specific AI optimizers.
For instance, if the Facebook agent finds an audience segment converting cheaply, it could tell the Google Ads agent via A2A to adjust search budgets accordingly. Your “meta-agent” orchestrates a symphony of channel-specific agents to maximize overall campaign performance, something that doesn’t happen across platforms today.
Real-Time Buying and Negotiation
Imagine an AI media buyer that not only auto-bids in exchanges but can actually negotiate deals.
With A2A, an advertiser’s agent might communicate with a publisher’s agent to secure a custom ad placement at a favorable price – essentially an AI-to-AI negotiation. This is like two bots emailing each other to finalize an insertion order while you supervise.
It could extend programmatic concepts into areas that still involve human negotiation (think upfronts or direct buys), making them faster and data-driven. You set the parameters (target audience, budget limits, KPIs), and the agents bargain and execute deals in seconds, across multiple publishers.
Campaign Troubleshooting and Support
Consider how many hours media planners spend monitoring campaigns and troubleshooting issues (under-delivery, creative fatigue, pacing problems).
In an agentic system, if something goes off track, your analytics agent (via MCP it’s monitoring performance data) can alert your buying agent, which then pings a creative agent to generate a new ad or a different call-to-action – all through A2A communication. They could even loop in a finance agent if the issue is overspending, to get budget approval for shifting funds.
In other words, agents can proactively handle the small “fixes” on their own, only escalating to you the bigger strategic questions or approvals.
Media.Monks Deploys Thousands of Agents
To illustrate how powerful this could be, Media.Monks experimented with an AI-driven media mix modeling tool called “Clarity” that used “thousands of AI agents” to simulate different allocation strategies; the best-performing overlaps of those simulations became the media plan.
That’s right – thousands of mini-AI planners each tried a tactic, and collectively they found an optimal mix much faster than a single team could. This kind of massive parallel experimentation is a glimpse of how planning might evolve. Agentic AI can swarm a problem with many specialized pieces of itself, whereas a human team does things one analysis at a time.
Faster, Adaptive, Holistic
In short, MCP and A2A could make media buying faster, more adaptive, and more holistic.
Plans and buys might adjust in near-real-time, across all channels in concert, driven by AIs that understand both the context (thanks to MCP pulling in data) and the goals, working together (thanks to A2A).
For media professionals, this doesn’t mean you sit back and press “go” on an AI (at least not yet!). It means your role shifts more toward strategy, oversight, and creative thinking, while the robotic parts of execution and number-crunching become highly automated.
What It Means for Media Agencies, Planners, and Buyers
How will these changes affect the people in the media business? Here are a few key implications.
Shifting Roles (Less Grunt Work, More Strategy)
As agentic AI takes over tasks like data gathering, initial planning drafts, and even intra-campaign optimizations, media teams can focus more on high-level strategy and creative decision-making.
Your job as a planner/buyer could become more about guiding the AI, setting the right objectives and constraints, and adding a human touch (e.g., client knowledge, brand nuance, creative ideas). In other words, you become a coach or pilot, rather than an assembly-line worker.
The mundane spreadsheet updates or platform toggling could diminish. This is similar to how GPS automation in cars lets humans focus on the destination and route strategy rather than every turn – you’re still in control, but assisted on the execution.
Need for New Skills
With AI deeply integrated, media professionals will need to be comfortable working alongside these agents.
Skills like prompt engineering (i.e. giving AI clear instructions), interpreting AI-driven insights, and oversight of autonomous systems will become important. It’s less about mastering the UI of every buying platform and more about mastering the orchestration of AI helpers.
Also, understanding data (analytics, attribution) remains critical – agentic AI will supply insights, but humans must verify and interpret them. Agencies might start training staff on how to supervise AI or even hiring AI strategists who specialize in tuning and monitoring these agent ecosystems.
Integration and Interoperability Mindset
Media agencies will likely push vendors and partners to adopt these open protocols.
If one tool embraces MCP/A2A and another doesn’t, the one that doesn’t could become a bottleneck. Expect to see RFPs and client questions around “does your system support agent-to-agent communication or open AI integration?” The industry might rally around these standards (much like they did for programmatic APIs) to avoid being left out.
Interoperability could become a selling point for tech platforms. For agencies, this means part of your job is ensuring your tech stack plays nice with AI protocols – maybe collaborating with IT more than before.
Ethics and Oversight
With great power comes great responsibility. If AI agents are automating decisions (e.g., shifting budgets, selecting audiences, negotiating deals), agencies must have guardrails.
There will be a need for governance policies: e.g., setting limits on spend changes an AI can make on its own, ensuring brand safety in any AI-initiated buys, and avoiding biased or ill-advised decisions. Humans stay “in the loop” to review and override as needed.
In practical terms, this might mean new approval dashboards where you see what the agents are planning to do and give a thumbs-up or down. Planners and buyers become more like editors or air-traffic controllers ensuring everything stays on course and within policy.
Efficiency and Scale
On the upside, if leveraged well, agentic AI could massively increase the efficiency and scale of media operations.
An agency team might handle more campaigns or more complexity than before because much of the heavy lifting is automated. This could level the playing field for smaller agencies to compete with larger ones, or conversely allow big agencies to manage even larger spends with the same human workforce.
It might also change billing models or client expectations – if AI enables faster turnarounds, clients might expect quicker results. Agencies will have to redefine their value proposition: less about manual labor, more about the strategic brainpower and creative oversight they provide in an AI-driven world.
Overall, media professionals should see agentic AI not as a replacement, but as an extension of the team.
Just as calculators didn’t replace accountants but changed how they work, AI agents can take on drudgery and complexity, if we adapt.
Those who learn to ride the wave will likely deliver better outcomes (and spend less late nights on tedious tasks); those who ignore it might find themselves outpaced by competitors using “AI co-pilots” for their campaigns.
Preparing for an Agentic AI Future (Practical Steps)
Change is coming, but it’s still early. Here are some practical steps to help you and your team prepare for the rise of MCP, A2A, and agentic AI in media buying:
Stay Informed and Educate Your Team
Start by learning more about these protocols and the concept of agentic AI.
Share articles or run a workshop internally to explain in plain language what MCP and A2A do. For instance, Anthropic’s blog post introducing MCP or Google’s A2A announcement are good primers.
Building awareness demystifies the tech and gets your team thinking proactively rather than fearfully.
Experiment with AI Tools (on a Small Scale)
Dip your toe in the water. Try out existing AI-driven products like those Agentic AI media products already on the market (Google Performance Max, Meta Advantage+, etc.). While they aren’t using MCP/A2A to talk to each other yet, they give you a feel for AI-managed campaigns.
You could also experiment with an AI assistant for planning – even something like using ChatGPT or Claude (Anthropic’s AI) to generate a draft plan or gather insights, manually feeding it data. This helps your team get comfortable with AI involvement in planning/buying tasks.
Audit and Organize Your Data and Tools
MCP works best if your data sources are accessible and well-organized.
Take stock of where your key planning inputs live – media systems, web analytics, research databases, past plan spreadsheets. Work with your IT or data team to ensure APIs or data connections are available. Breaking down data silos and updating data hygiene now will pay off later, because you’ll be ready to plug into an MCP-like ecosystem where the AI can draw from all those sources.
Similarly, ensure your current tools are up-to-date and see if your vendors are planning support for agentic integrations.
💡Cheat code: Bionic for Agencies automatically centralizes, standardizes, and organizes all your media data.
Engage with Tech Partners
Ask your software vendors (for planning, buying, analytics) about their roadmap for AI integration. Are they adopting open standards like MCP and A2A?
For example, some media software providers are already enabling planning of AI-driven placements (Bionic, for instance, added an “AI Agent” channel for incorporating agentic AI products into media plans) – this is a sign they’re forward-looking.
Joining beta programs or pilot projects with vendors can give you hands-on experience. If a vendor has no answers on AI interoperability, that’s a red flag; you might prioritize partners who are embracing these innovations.
Develop AI Governance Policies
It’s easier to build trust in AI if you set clear rules for its use. Start drafting guidelines for how your team will use AI in media buying.
For example, decide what types of decisions AI can make autonomously vs. what needs human approval. Outline a process for reviewing AI-generated plans or optimizations (maybe a weekly review meeting of AI suggestions).
Think about brand safety and ethics: if you had an agent negotiating deals, what boundaries must it respect? Having a framework in place will make it smoother to integrate AI agents when they become available, because everyone will know the rules of engagement.
Train and Upskill Staff
Identify the skills your team will need in an AI-enhanced workflow and start building them. This could mean training on data analysis (so staff can validate AI findings), learning prompt design for interacting with AI, or even basic scripting to tweak AI tools.
Encourage a culture of curiosity and continuous learning – perhaps assign interested team members to become your “AI champions” who follow the latest developments and report back. The goal is to make AI a familiar ally, not an alien threat.
When people understand it, they’re more likely to leverage it effectively.
Start Small, Then Scale
When agentic AI capabilities (like tools supporting MCP/A2A) start becoming available, pilot them on a small project or one campaign. Treat it as an experiment; gather results and lessons.
For instance, maybe in 6-12 months you get access to an AI that can use MCP to pull data and suggest a plan – try it for a niche client or as a shadow plan alongside your normal process. Measure the outcomes and efficiency. Use those insights to make a broader adoption plan.
This phased approach lets you build confidence and proof points before fully rolling out new workflows agency-wide.
Foster a Collaborative Culture (Human + AI)
Finally, prepare your team culturally to work with AI.
Emphasize that AI agents are like teammates or assistants, not job replacements. Encourage people to see the value in offloading grunt work and focusing on higher-value tasks. Celebrate wins where AI helped (e.g., “our AI suggestion improved CTR by 20% – great oversight by Jane in tuning that”). A collaborative mindset will ease the transition. After all, the best outcomes will come from human creativity plus AI efficiency working together.
By taking these steps, you’ll be positioning your agency or team not just to survive the coming changes, but to thrive. The media landscape is only getting more complex – new channels, more data, faster pivots – and agentic AI could be the key to managing that complexity at scale.
Protocols Like MCP and A2A are Kind of a Big Deal
MCP and A2A might be technical acronyms today, but they herald a future where our media buying tools are smarter, more connected, and more autonomous.
Just as digital buying upended traditional methods, agentic AI could redefine the playbook once again.
The good news is that by understanding these concepts early and preparing our people and processes, you can harness the benefits (and avoid the pitfalls) of this technology.
It’s an exciting time to be in media – we’re not handing the keys to robots and heading to the beach; we’re equipping ourselves with powerful copilots for the journey. And that means potentially better results for clients, less tedious work for teams, and more time to focus on strategy and creative ideas that truly move the needle.
The post How MCP and A2A Are Poised to Disrupt Media Buying first appeared on Bionic Advertising Systems.
News and Articles | Bionic Advertising Systems
