I have spent 20 years building ad tech infrastructure, from early native advertising platforms to real-time bidding systems to first-party data products. Across every one of those chapters, I have watched the same pattern: the companies that waited for a new standard to "stabilize" before engaging were the ones that got left behind. 

That pattern is playing out again with the Ad Context Protocol. 

AdCP is the first serious attempt to standardize how AI agents communicate across the entire advertising ecosystem. Think of it as what OpenRTB did for programmatic auctions but applied to the full media cycle. For two decades, real-time bidding was the only machine-to-machine standard we had. Everything else, from creative briefing to audience segmentation to the sales process, still ran on emails, phone calls, and bespoke integrations. AdCP changes that. 

The protocol is moving fast. AdCP 3 is rolling out now, expanding beyond media transactions to cover creative workflows, audience data activation, and AI publisher signals. 

I understand the instinct of letting it settle. 

But here is the paradox I keep explaining to my own teams: yes, if you wait six months, you could probably build your integration in three days using spec-driven engineering. But you will not be in the position to do it, because you will have missed the context, the iteration, the learning that comes from building alongside the standard as it takes shape. The companies engaging now are writing the playbook. Everyone else will be reading it. 

The case for early engagement goes beyond the first-mover advantage. Consider what AdCP means for the open ecosystem. Today, if you are not one of the major walled gardens, competing at their scale requires custom integration with hundreds of vendors. Switching your creative partner means migrating your entire business intelligence layer. But when every creative platform, every DMP, every publisher follows the same protocol, those lock-in dynamics dissolve. One vendor for CTV creatives, another for display, a third for everything else, all without rebuilding integrations each time. For anyone operating outside the walled gardens, that interoperability is not a nice-to-have. 

It is how the open web stays competitive. 

I will be honest about where things stand; not many agentic dollars are flowing through AdCP today. In some markets, we are still looking for the first live campaign. The hype is ahead of the revenue. But that gap is precisely the point. The founding members - like my company Affinity - are not just implementing a finished protocol. They are shaping for what it becomes. Early participation buys influence, not just capability. 

There is one risk I think the industry is underestimating: observability. When programmatic was new and something went wrong, say a campaign accidentally running globally with no budget cap -you could log into a UI and fix it. In an agentic world, mistakes propagate faster and at greater scale, often without a dashboard to catch them. AdCP has answers to this. But only if you're paying attention to it. Companies building agentic systems without prioritizing human legibility are building on borrowed time. And this is a challenge you can only understand by working with technology, not by watching from the sidelines. 

The protocol will keep changing. That is not a reason to wait. It is THE reason to start. Every major shift in this industry, from programmatic to header bidding to privacy reckoning, rewarded the people who showed up early and adapted as they went. 

The cost of building now is iteration. The cost of waiting is irrelevant.