• by Dave Morgan , Featured Contributor, Yesterday

In law school, you learn about the principal-agent relationship: the rules that govern the responsibilities and liabilities when one party acts on behalf of another. These used to be big issues in the U.S. media business, but that was before the holding companies largely pivoted to principal-based trading to pay their bills.

Now, however, with many marketers and agencies talking about using agentic AI to automate and optimize their media planning and buying tasks -- or, more properly, to help them best achieve their media objectives -- the notion of principal-agency relationship is squarely back on center stage.

And key questions for all will be: “Who is this AI agent truly working for?” Is it working for me? Is it working for the company that created and licensed it? Or,might it be working for an undisclosed party operating in the background?

If the history of ad tech is any guide, we know that unless you built it yourself, your agentic AI media buying bot is unlikely to be working for you, and much more likely to be working for the company that you license it from, or some other other company operating in the background secretly subsidizing it -- just like the SDKs in so many apps on smartphones and smart TVs.

Sad, but true.

Okay, so you want to learn what you can do to better negotiate with one of these thousands of new upstarts big and small offering to build you your very own agentic AI “agent.” Here are some of the principal-agency issues you should you be on the lookout for:

Divergence of interests. If your agent’s interests are not fully aligned with yours, you’d better clearly understand where and how it diverges, and know what this means for your interests.

Information asymmetry. If your agent knows more about what it is doing, and not doing, than you do, you will be at risk. Are you contractually guaranteed transparency? Can you create a meaningful “trust but verify” structure in your relationships?

Agency costs. Do you truly know all of your “costs” of using the agent, which include monies, discounts, etc. that they take from suppliers and service companies that would otherwise flow to you?

Accountability and governance. Is there true accountability in your relationship? Do you have real transparency into how the agent operates, what governs it, and how you ensure that you can have a “trust but verify” mechanism?

These are only places to start. Truly managing agentic AI agents and ensuring that they work for you -- and only you -- will be easy to say and hard to do.

Thus, my question to you: Do you trust today’s arbitraging ad-tech industry to be your agentic AI?