By Paul Frick
When developing a communications strategy, the first question I ask is: who are we speaking to?
The answer shapes almost everything else that follows: know the audience and you can set the message, choose the channels, and decide the content you need. Knowing the who tells me the how.
But lately I’ve started to consider another layer of audience. As AI becomes more typical of how people find information, I have to ask: are AI models themselves an audience?
My conclusion is yes, communicators need to plan for AI, but not because AI is an “audience.” AI has become something different, and more influential, than a single audience: it is an interpreter that stands between you and many of the audiences you are trying to reach. Ignoring it would be risky.
The AI Intermediary Problem
Communications strategy has always been shaped by whoever stands between organizations and their audiences. For decades, the media played intermediary with journalists and editors deciding which stories reached the public and how they were framed.
But now AI is changing the intermediary. Today, when a policymaker, client, or a reporter on deadline wants to learn about you, there’s a pretty good chance that their first stop is not your homepage, but an AI model. Before they encounter your own content, they will encounter the model’s interpretation of what it knows about you.
So by the time your audience reaches your content (if they reach it at all) your story may have been interpreted and summarized by an AI system you do not control and placed in a context you didn’t choose.
AI is increasingly replacing traditional search. A Pew Research study found that 58% of participants encountered at least one AI summary while searching over a one-month period in 2025. When an AI summary appeared, users clicked a conventional search result only about half as often.
AVOQ’s annual 2026 Media Habits research points to the trend: 96% of policymakers reported that they are using AI, and half use it daily. But more consequential for communicators – 78% told us they plan to rely on AI even more in the year ahead.
So, we’ve added another intermediary, if not quite an audience, and it behaves very differently. It does not deliver a consistent interpretation. The same question can produce a different response as models incorporate new information, draw different connections, coverage, or respond to a slightly different prompt.
That makes the work harder. By the time you respond to one characterization other versions may have surfaced. Different models or even different phrasing of the same prompt can change the answer.
AI models also cannot be engaged in the way we’re used to engaging with the media. You can call a journalist or write to an editor, and over time build a relationship. There’s no email address or phone number for an AI model. The tools we spent decades sharpening, including access, relationships, and the well-timed call, don’t transfer neatly.
How to Respond?
The job of the communicator has not changed. The path to our audience has.
If the first interpretation of your organization increasingly comes from AI, communicators need to plan for that reality. The good news is that AI’s interpretation can be influenced, but only if you approach it deliberately.
Like it or not, AI now belongs in the answer to my first question: who are we speaking to?
Check back for Part 2 of this series, I’ll talk with Mimi Tan, Avoq’s VP of user experience and content strategy, about what organizations can do to shape how AI interprets and presents their story. It’s an ongoing discipline, and one worth the investment.
