Have you ever wondered if your comms agency—whether that’s PR or content or both—is using AI? And then wondered, should I even care?
The answer is pretty clear: Yes, your agency is using AI, and yes, you should care.
And if companies should care whether their agencies use AI, then it’s important that we make it clear how Propllr is—and isn’t—using AI (fwiw all of these em-dashes are mine, not AI’s!).
Because I’ve had a bit of writer’s block of late, I asked ChatGPT to pretend it was a prospective client asking me questions on this topic….
1. How do you use AI today?
My take on AI use is this: Are we good or bad at doing this thing that AI can do? We are great at (and love doing) content planning, media targeting, writing, interviewing, strategy, etc., so we won’t use AI there except around the edges as we refine, validate, or edit. Where AI makes sense is on work where one’s talent or ability to grind is less important.
So we use AI for quickly learning about a space we do not yet know, for being an idea partner, and for validating ideas. In all of these cases, we think it makes our work more in tune with our clients and their markets.
That said, if we were to combine all of our AI efforts, I’d guess that AI hasn’t (yet) made us any more productive. Sure, some things might get done more quickly with AI, but many AI efforts become wild goose chases with little of value to show for them.
2. What platforms do you use, and how do you vet them?
For client work we only use paid versions of “enterprise-grade” AI platforms in which our work is kept private. This includes ChatGPT, Claude, and Perplexity as well as vertical platforms like MuckRack, which often have AI-powered tools. Are we taking their word that they’re not using our work to train their models? Yes.
3. Can you show an example of how AI has improved speed or quality for a client?
We don’t track AI “wins,” but I would guess the areas where AI improved speed or quality for a client would be editing our work, validating ideas, and researching new topics.
That said, we did have a client looking for speaking opportunities, but they were rather new in their space and had just limited traction at the time. This meant that the big conferences were unlikely when it came to speaking opportunities. We crafted a prompt that explained that context and asked for off-the-beaten-path trade shows where we might have success placing them. It surfaced a few small but good events, and we landed our client a speaking spot on one of them.
4. How do you ensure that AI-generated content meets our brand standards and voice?
We don’t use AI to write content for our clients, but we can’t just assume a hand-crafted blog post will stand out. Over time, as more and more companies turn to AI for content, the output will get better and harder to tell robot from writer. It’s a daunting challenge, but the reality is that it’s more important than ever that we help our clients develop distinctive voices so that they stand out as leaders.
To stand out today, we are pushing clients to provide deeper insights and more unique perspectives, using concrete language and plain english, and incorporating quotes or insights from their customers, partners, influencers, etc.
5. How do you protect against factual inaccuracies or hallucinations?
We see AI as a precocious Wikipedia: It thinks it’s smarter than it is but doesn’t have as much oversight. Trust by verify, as they say. And again, we’re not using it for output, so that in and of itself limits risks in this area.
6. How do you stay up to date on AI’s impact on communications, media, and search?
If you’re in comms, it’s hard not to stay up to date with this in the course of doing business. Really, it’s lots of reading blog posts and newsletters, lots of listening to podcasts, and lots of talking to clients and peers. Rand Fishkin’s thought leadership is driving a lot of our thinking in this area. Highly recommended for all marketers.
7. How does AI support your strategic recommendations?
We’ve not implemented anything like this company-wide, but AI is well suited to improve strategy. It can be an “expert” to turn to for validating ideas, spotting gaps in our thinking, and poking holes in our beliefs. It’s like getting challenged by a supportive colleague who wants nothing but the best for you (although AI’s desire to please can be problematic).
8. What’s your philosophy on balancing human creativity with a machine’s efficiency?
If you’re good at something, just keep doing it, and use AI around the edges. If you’re not good at something or need to know more about an unfamiliar subject, ask AI to help you through it. And if you’re neither good nor bad at something—you just don’t want to do it and no one is paying you to do it—feel free to ask AI.
9. How do you train your team to use AI effectively and ethically?
Right now, with everything so new, it’s not so much about training folks to be effective on AI as it is encouraging them to try new things with AI and to share their experiences, both good and bad. And yes, the ethics of AI are important. On the client side, we need to protect their information, and on the energy and environment side, we want to keep an eye on the impact of our actions.
10. What role does AI play in media targeting?
We pay an exorbitant amount to MuckRack, and that’s where a lot of our media list building and targeting decisions occur (AI has taken root there, firmly). But outside of MuckRack’s AI, we certainly test the dominant AI platforms, with wildly variable results.
For example of where AI falls short in this area, we might ask ChatGPT for a list of 20 podcasts that serve a particular audience and be excited by the results. But then, once we start looking through the names AI gave us, we’ll find half of the podcasts aren’t real and many of the rest are no longer active. AI has been a big tease on a lot of this stuff.
11. How do you ensure AI doesn’t introduce bias in our messaging?
The techno-optimist in me is more excited by the idea that with the right AI prompt, we can actually spot and fix bias in the work we humans create. It’s still early in figuring this out, but there’s a lot of potential there.
12. How does AI affect how you think about SEO and visibility?
There’s a lot being written right now on GEO, or generative engine optimization, which is how to get your content positioned so that LLMs spotlight it in their answers. Many of the recommendations seem similar to SEO, but there are a number of other things companies need to be doing with schema and other technical issues.
It feels like the early SEO days, where folks would crack Google’s algorithm and get on the first page of search results, but then see an updated algorithm push them back down. For now, we’re staying on top of GEO and ready to make recommendations, but so long as we continue doing great work across content and PR, I think we’re setting our clients up for success.
If you want to stay on top of the conversations, make sure you bookmark the SparkToro and the Ahrefs blogs.
13. Do you disclose to clients when content or strategy recommendations were AI-assisted?
We’ve made no recommendations to clients that were simply copy-pasted from AI. That said, I imagine that at some point in the not too distant future, we will have client-specific GPTs where we can feed in new information and ask it to generate some new ideas, review data, etc. When we do, we would definitely collaborate with our clients before utilizing them.
14. How are you adapting your services in response to how AI is changing market dynamics?
Our goal is to help B2B technology companies build trust, and we do that with content and PR work that focuses on building credibility, establishing authority, boosting community, and demonstrating momentum. Because AI is leading to an explosion of content and noise, I think we will increase emphasis on authority and community, two very human areas where AI falls short.
15. Did AI help you write this blog post?
As noted in the intro, I used ChatGPT to help me come up with the questions and to find relevant research to cite in my answers. It did pretty well with the questions, but not as good with the research.
Looking ahead at Propllr and AI
Propllr’s approach to AI can best be summarized as trust-first, human-led, experimental, and optimistically skeptical, and I don’t imagine that will change. That said, I will revisit this Q&A as AI evolves.
And if you have questions that aren’t reflected above, get in touch!
