Eight short episodes on how AI agents are starting to plan, buy and optimise media — and the open protocol underneath it. The seven-part core runs in under 30 minutes, plus a bonus on the people behind the agents.
In this series
The gateway. What actually changes when machines can buy media on your behalf — and how agents that transact differ from the AI-assisted tools you already use.
The Ad Context Protocol, in plain terms: one open standard that lets any buyer agent talk to any seller agent. How it sits above OpenRTB rather than replacing it, and who governs it.
A real end-to-end walkthrough: a plain-language brief goes out, comparable proposals come back, delivery reports through one view — and the honest answer to whether real money has actually moved.
Governance as protocol, not policy. The binding check that runs before every buy, the escalation threshold you set, and the signed audit trail that proves what was requested, approved and by whom.
Protocol versus platform. Why going agentic doesn't mean moving your brand memory and first-party data into someone else's operating system — and how activation can happen without the raw records changing hands.
Shared plumbing isn't shared strategy. Why standard rails don't commoditise your advantage, what happens to prices, and how the friction that protected the walled gardens starts to fall away.
A practical playbook in three moves — ask and scope, run it governed, judge and decide. One campaign, one owner, one spend cap you can defend, and a baseline to measure against.
The people question. Who the "agent managers" really are, who trains the agent (you do), why approval rights concentrate with experience, and which media jobs still need humans in five years.
This series is my attempt to demystify agentic advertising for the people who'll actually run it. If it's useful, or if you want to compare notes, come find me on LinkedIn.