Everyone's talking about LLMs replacing SEO. But something bigger is happening that most people aren't seeing yet. OpenAI isn't just building a search engine. They're building an entirely new ad network, and if history teaches us anything, the agencies that move on this early will take a huge step in front of the competition. Remember early Google Ads? The companies that moved first made millions off cheap clicks before everyone else figured it out. We're at that moment again.
It's not exactly a secret. OpenAI is already hiring for its advertising platform. But here's the bigger picture: LLM-generated search traffic is projected to surpass Google search traffic by 2028. That's a fundamental change in how people find information, compare options, and make decisions. When platforms reach that scale, they monetize with advertising. Google did it. Facebook did it. OpenAI will too.
Today's ads fall into two categories: keyword-based ads on Google, where you bid for search terms and drive people to your website, and creative-based ads on social platforms where you show videos or images and send people to a webpage or form. OpenAI's ad network won't work like either of those. The new format will be conversational and interactive. The ad unit itself will live inside the AI conversation, not as a link to click or a banner to ignore.
Here's what that could look like for insurance: A consumer asks ChatGPT, "Help me lower my monthly expenses." GPT analyzes their spending and identifies that they're paying too much for auto insurance. It then triggers a call to the ad network, essentially saying, "I have someone who needs an auto quote. Who wants to bid on this?" The winning agency surfaces a live quote directly inside the chat. The entire buying journey happens without the consumer ever leaving the conversation.
This isn't theoretical for us. Back in 2016, Facebook launched messenger ads. These were conversational ad units inside Facebook Messenger, and Mav was one of the first companies to get access. We built a use case with one of the largest insurance companies in the world. At F8 that year, Facebook showcased our work as one of the most successful implementations of the format.
The difference now? The channel is bigger. The context is smarter. And the stakes are higher. LLMs understand intent better than any previous platform. They're already in the middle of helping people make decisions. When ads move into those moments, the conversion potential is massive.
To participate in this ad network, you'll likely need an MCP server. That stands for Model Context Protocol. Think of it like an API for a mini bot that does one specific thing. For insurance, it would be an API that delivers quotes. The MCP is what plugs your offering into the LLM so it can surface your service at the right moment. OpenAI will probably make this easier later with templates and no-code interfaces. But that won't be phase one.
Phase one will require technical setup. And the opportunity cost of waiting for the easy button? Huge. The agencies that get their MCP capability ready now will have first-mover advantage when this goes live. The ones that wait for the simplified version will be competing in an already crowded space.
If you're an agency that relies heavily on lead flow, things are about to change for you. Traditional ads alone won't cut it anymore. The ad unit itself will need to deliver real value in real time. This changes how leads are generated, how intent is captured, and how quotes are delivered. Right now, you're probably buying leads from vendors or running Facebook and Google campaigns that send people to landing pages. That model worked because you could optimize for cost per lead and manage the math from there. But when the ad unit is the quote itself, the game changes. You're not buying leads anymore. You're bidding for the right to deliver value inside the conversation. The agencies that can do that well, and do it fast, will own the next wave of AI-driven traffic.
Start by understanding what an MCP is and how it works. You don't need to build one yourself right now, but you should know what it does and why it matters. Then get your MCP capability ready before the no-code version arrives. This is like the early days of Google Ads. The first movers captured wealth because they understood the mechanics before the masses showed up.
If you're already a Mav customer, you'll have day-one access to this capability when it goes live. We're building for this shift right now. If you're not, it's worth getting in touch to talk through what MCP readiness looks like for your agency. Because once this launches, the learning curve will be steep and the window for early advantage will close fast.
Everything here is still forward-looking. OpenAI hasn't flipped the switch yet. But the trajectory is clear. When ad units move into AI conversations, every agency's acquisition playbook changes. The platforms change. The formats change. The economics change. The question isn't whether this will happen. It's whether you'll be ready when it does.
Watch our CEO Matthew Black explain these concepts in-depth in his newest video.
