In this episode, Chip and Gini discuss the impact of AI on small agencies, focusing on the high expectations and possible disappointments it poses. They reference a recent article from The Atlantic, which highlights a study showing that AI can sometimes decrease efficiency.

They caution against overhauling business models based solely on AI’s current capabilities, stressing that while AI can assist with tasks and improve efficiency, it cannot fully replace human judgment and creativity.

The conversation extends to the challenges of integrating AI without sacrificing the development of new talent and ensuring that the evolving role of AI adds value rather than causing disruption.

The following is a computer-generated transcript. Please listen to the audio to confirm accuracy.

Chip Griffin: Hello and welcome to another episode of the Agency Leadership Podcast. I’m Chip Griffin.

Gini Dietrich: And I am Gini Dietrich.

Chip Griffin: And Gini, I think, you know, I think AI is gonna solve all of my problems. We’ve said this before, but I, I really, I’m gonna change my whole business model around it.

Gini Dietrich: I think you should fire everybody.

You could fire me as your co-host. AI can just do it for you.

Chip Griffin: You know, I mean, I wonder if the AI would talk back less.

Gini Dietrich: Probably, although if I program it, I’ll program it for you so that it, it talks back a sufficient amount.

Chip Griffin: Excellent, excellent. Just, just what I need.

Gini Dietrich: You’re welcome.

Chip Griffin: Oh, at least, you know, maybe, maybe the AI wouldn’t yawn at me like you did just before we started recording, so

Gini Dietrich: I shouldn’t even be tired either.

I got nine hours of sleep last night. I don’t know what my problem is,

Chip Griffin: I assume you were expressing your boredom with me.

Gini Dietrich: No. Prior to the start of the show. No.

Chip Griffin: I mean, usually it takes a couple minutes for most of our listeners to get that bored with me. So

Gini Dietrich: no, I’m not bored with you.

Chip Griffin: Okay.

Gini Dietrich: Just sleepy.

Chip Griffin: Just sleepy. Well, I’m with you on that. Alright. But we are gonna talk about AI and in particular, there was a good article in the Atlantic recently, that is consistent with some of the things that I’ve been saying and talking with clients and, and I think it’s, it’s helpful to think about. And that is whether or not there is an AI bubble and what it means.

And it’s not, in my view, it’s not so much about a macro AI bubble, sort of like we think about, you know, the, the internet bubble in the early 2000s and, and I think there is certainly some of that, but that’s really more about investing in some of the businesses and infrastructure and that kind of thing.

But I think for small agencies, there is the potential for an AI bubble that comes about because we all get too focused on either what kind of threat AI is or what kind of opportunity AI is. And we’ve said before on this, we’ve talked about AI a lot, so I, I was a little reluctant to come back to it, but I think that thinking about it in in these terms may be helpful.

Because it’s not, it’s not about AI being either the panacea or evil, it’s that we may be adjusting our models too much too quickly because some of these things may not quite be there. And in particular, the Atlantic had a great example and it was from a study that was done with some software developers. And the software developers were given AI to help them do their jobs. And the purpose of the study was to try to figure out was it a 20% efficiency improvement, 40%, 60%, what was it? And ultimately what they discovered was it was actually, it decreased their efficiency.

Gini Dietrich: Mm-hmm. Mm-hmm.

Chip Griffin: Because part of the challenge with AI, and I, I’m sure many of us have seen this, is when it works well, God, it works great.

It just, and, and we’re impressed and we’re, we sit there and say, that’s amazing. But we, we don’t factor in the times it doesn’t work and the times we have to argue with it and fight. Right? And so if, if we’re go, and I, I mean I do this a lot with, I mean, maybe it shows something about me that I argue with the computer.

I was trying out Gamma recently, which is a, a service that lets you build PowerPoint decks effectively. And I threw something in, just kind of randomly grabbed an outline, threw it in, and it actually did a really nice job of creating the first thing that I sent to it. And I’m like, that’s great. I’m like, okay, well can you make these modifications to it?

And it just wouldn’t do it. It just, it, it kept, you know, I, I said, can we put, you know, my logo at the bottom of the slide, you know, in the footer. And so it puts it in, it’s like a quarter of the screen. I’m like, no, no. I said, can you make it smaller,

Gini Dietrich: same size, just move it.

Chip Griffin: And it’s like, great, that’s a great idea.

Let’s make it smaller. And it gives it to me the exact same size. Oh. And it makes it completely unusable because obviously you don’t really want your company logo to take up a quarter of the slide. That’s just dumb. So I, I think we need to, to think about all of the effort that we’re putting into AI and factor that in when we’re thinking about our own efficiencies, because I’m afraid that some people may be changing their models too much, changing their pricing too much. Because they’re looking at the successful implementation of AI and not factoring in all of the challenges along the way. At least as it exists in September of 2025.

Gini Dietrich: Gartner has their hype cycle, right? And so it looks at things from that perspective. And, you know, we get really excited. We’re human beings. We get really excited. There’s something. Coming. There’s like all this like, oh, that’s scary. We’re not gonna do anything. Then we get really excited about it. Then it dips down and then it comes back up before it’s like scalable and, and something that we use.

The internet was one, email was one, social media was another, those kinds of things. Right. Well, in the hype cycle, AI is now in the trough of disappointment. So it’s in that section of the cycle where it’s like, oh. Maybe this isn’t gonna work so well. And you have companies like Salesforce who just laid off 30,000 people.

Because they think AI is going to replace those human beings. And what they’re going to find in the trough of disappointment is that that’s not the case. And I think that we have to think about that. As you know, we’re certainly, I’m not laying off 30,000 people. You’re not laying off 30,000 people. I would imagine most of our listeners are not laying off 30,000 people.

But you have to think about it from that kind of perspective of AI is great for some things. It cannot replace human beings yet. So when you think about how you’re using it, those are the kinds of things you have to consider. One of the things I say to my team all the time, I, I’ll get something from a team member and I’ll be like, did you use AI for this?

And my favorite is when they’re like, oh, just to outline. I’m like, really? ’cause on the third page, the sixth bullet point, it’s definitely AI. And I know that because you didn’t, it’s, it’s talking to you, right? And you didn’t edit it. So it can’t replace people yet. And I think though, that’s what I continue to tell my team is you can absolutely use it, but you have to use your brain and you have to edit the work that you’re getting because as a first draft, it’s crap.

It’s not good. So we have to like take all of this with a grain of salt and make sure that we’re using it effectively.

Chip Griffin: Yeah. And, and I think that, you know, part of the challenge is if, if we say, okay, well because of this we can now, you know, churn out social or blog posts or graphics or whatever that much faster and all that and, we’re building our models, we’re building our staffing models around it. We’re building our pricing models around it. It becomes a problem. Yeah. Yep. When we realize that it’s not quite as efficient, it’s still good. It’s still helpful. We still ought to be using it and some of this stuff will continue to get better.

Most of it will continue to get better, and so at some point it will get there. It’s sort of like during that, that first wave of the internet bubble in the early 2000s, there were plenty of businesses that came out, got a lot of attention, and then failed miserably only to see 10 or 15 years later an almost identical business come up and blossom because now it was time for it.

The technology was there, the market was there, and so, so some of these things, it’s not that they’re not a good idea, it’s just that they’re not quite fully baked yet. And as, as particularly as small businesses, we need to be careful about how heavily we lean into some of these things without understanding what the risk of that is and, and understanding whether it is truly everything that we believe it is in our, in our happiest moment.

And, and, you know, with the troughs of disappointment and the, the peak, I forget what the, what the peak is called, but you know, the ups and downs, we’re all gonna feel those. But, but when we’re building business models and strategies for our agencies, we need to think about that in a little bit more of a level line as opposed to to riding the waves or the dips too much because that’s where we start to make bad decisions.

Gini Dietrich: Yeah, and I think we, you see this on the, on the web all the time, right? You have people who jump onto the next wave, hoping that they’re gonna get that upswing of popularity, success, whatever it happens to be before it goes down.

And then you have, you see ’em jump on the next one and the next one and the next one. First of all, that’s a crappy way to live. Like, ugh. The constant like yin and yang and ups and downs and no agency owner should live like that. But secondly, you have to really think about what are we strategically trying to do?

And I will tell you, I, I mean, I’m sure it’s clear, but I love AI. I love it. I think it has made me more efficient, it’s made me more productive. You know what it’s done more than anything? It’s allowed me to stop procrastinating. Because when I have something really challenging, I tend, in the past, I tend to procrastinate it, put it off.

It’s a big project. It’s, I know it’s gonna take me time to get started, and so I, I put it off. I put it off. I put it off. I’m also highly deadline driven, so if you tell me it’s due in 12 hours, it’ll get done. If you tell me it’s due in four weeks. I’ll do it the last 12 hours. Right. So that’s just how I, my personality.

So what it has done is it has allowed me to say, okay, this is a huge project. I’m, I’m gonna take 15 minutes just to outline the steps so that I can get it done more effectively. I don’t procrastinate anymore because that allows me to at least get it into my workflow and get things accomplished. That’s where it’s made me most, most efficient.

So I think we have to think about those kinds of things, like how are we bringing it into our workflow to make us more productive and make us more efficient, but not replace the thinking or the creativity.

Chip Griffin: Right. And I think the other thing that we need to keep in mind is that in uses like that, where it’s acting as our assistant, I think has a lot more utility, a lot more staying power. Mm-hmm. And will just continue to improve as, as we learn how to use the technology more effectively. And as the technology continues to evolve, I think the bigger challenge comes for agencies when they see this as something that can help them do things at scale.

Yes. Right. So it’s not, it’s not assisting an individual, making them more effective, efficient, et cetera. But instead it is using it, as some companies are, to replace large numbers of human beings.

Gini Dietrich: Yep.

Chip Griffin: And I think that’s, it’s a risk for multiple reasons. One is the, you know, whether or not it is truly as good as you think it is and, and can do things without that level of supervision, intervention, et cetera, that we’re doing when we’re using it as an assistant.

But we also need to keep in mind that as we all get sucked into this, it is incredibly likely, if not certain, that all of the providers of AI are going to start ratcheting up the prices. Sure. And so what looks today like a huge cost savings likely may not be. In just a few short years. And we’ve seen this before, we saw this in the early days of internet ads, for example, search ads.

It used to be that you could, you could buy ads on, on Google search super cheap. Now, not so much. And it didn’t take all that long before it got to a place where it was so costly that it was starting to, to become much more similar to other advertising media. Right.

And so you, you need to keep in mind that it may not ever get fully to the price of, you know, the, the same productivity of a full-time equivalent, cost the same in the end, but it’s gonna be a lot more than it is today. So if, if you see it today as an 80% savings, it may only be a 40% savings when all is said and done. And is that still okay?

Would you still have uprooted your whole business model for that or not? And I can’t answer that for you, right? But I think you need to think those things through, before you invest too heavily. What’s that?

Gini Dietrich: It depends.

Chip Griffin: It depends. Everything depends.

Gini Dietrich: Yeah. I think you’re absolutely right.

Like it’s just, you know what, there’s been a conversation both in a private Facebook group I’m in and in the Spin Sucks community of late, of how people are not hiring new college graduates anymore because it kind of does replace them. Like it does replace the research, the media list development, the initial pitches, things like that that you would have, you know, an intern or a young prof, a brand spanking new young professional do.

So how do we, how do we continue that? But then how do we hire young professionals? ’cause we can’t not hire a brand new college graduate. Right. How do we bring them into the workforce and what are we, how are we teaching them? Are we teaching them to become AI prompt engineers? Are we teaching them to become better editors?

Like what are we teaching them to do? Because AI can do a lot of that really basic stuff that an intern would would normally do.

Chip Griffin: Absolutely. And, and I think that that, that becomes then a, an even bigger challenge for small agencies because small agencies have always struggled with the idea, do we hire seasoned professionals who are more expensive? Or do we hire more junior people who we can use more cost effectively?

And this does tend to skew it more in, in terms of let’s, you know, let’s hire those more seasoned, more expensive people and we’ll rely on the AI for all the lower level stuff. But that may be a challenging business model to truly make work, in part because now you’ve got senior people who then have to, to manage in a much more hands-on way with the AI.

A lot of that more basic stuff. And I would argue that it might be more worth looking at can you use the AI to improve the, the quality and output of more junior people, right? If, if you’re, if they’ve, if the junior people have an assistant, could they accelerate their own growth more quickly? And I don’t have the answer to whether that’s true or not, but it’s something that I would be interested in looking at if I were trying to scale up a small business as opposed to saying, okay, all the grunt work will be done by AI and we’ll hire just senior people, because that’s a very difficult business model to price correctly.

Gini Dietrich: I also think it’s dangerous. I think it’s dangerous to hire as, as a senior professional, I think it’s dangerous to hire only, you know, seasoned professionals because you now you’re losing out on, you know, creativity, youth, you know, like all of the, the sort of like rosy colored glasses kinds of things that help a business grow.

I think that’s, I think that’s really dangerous as well. You need, you need a balance.

Chip Griffin: Yeah. And you wanna develop future leaders for your own. Yep. Business. You know, obviously it, you know, I don’t necessarily buy into the notion that we need to do it for the good of the PR and marketing community and, you know, we gotta give them opportunities because otherwise we won’t have them in the future.

It, it really does need to be more selfish for your own business. But you can develop them within your own business so that you know, over the next, you know, four to six years, they become that seasoned professional who learns your way of doing things as an agency and can contribute in a more meaningful way to your own business.

Gini Dietrich: I think it’s, I, like I said, I think it’s dangerous to, to rely on that. And to come full circle, it’s dangerous to rely on AI to replace human beings because that’s just, while it’s tempting, it’s just not where we are yet.

Chip Griffin: No, I mean, look, I think there are, there are certain things that can be replaced by AI. So I’m not, I’m not sticking my head in the sand and saying, don’t, you know, don’t reduce any headcount. I, I think for most small agencies, probably AI is not gonna enable you to reduce headcount. It’s, it’s difficult for me to, you know, all the agencies that I’ve advised and looked at over the years, it’s hard to think of very many who could actually save even one FTE on the backs of what AI can do for them today.

Yep, yep. It certainly, it certainly can make them more efficient. It certainly can open additional doors, could free up some time. But I, I have a very difficult time seeing any where it’s a one for one replacement. But there are certainly other industries, other kinds of business where it absolutely can help to reduce the headcount, but maybe not completely replace all of the headcount in a particular function area.

Gini Dietrich: Well, to come back to the Atlantic article that you mentioned at the start, you know, it showed that it wasn’t, and you would think, you would think that AI would be really good at a developer’s job. At creating, you know, being able to do that because it’s not, there’s no creativity to it, right? And so you would think it would be really great at that.

And it found that it wasn’t. It found that they were actually spending more time directing the AI and getting the information, getting the right kinds of things out of the AI that they needed than to just do it themselves. So you, we have to be really careful here and, and like I said, I’m a huge, huge fan.

I love it. But it’s also not replacing any humans from my perspective.

Chip Griffin: And look, I mean it ultimately, even though the, the models are very good at doing the coding for programmers, it’s also, it’s a double-edged sword because for an op-ed or something, 80% is probably good enough. For code? 80% isn’t good enough.

You gotta get to a hundred percent because any bugs in there are gonna be a problem.

Gini Dietrich: Yes, you do.

Chip Griffin: And, and as someone who has written computer code before, I can tell you that, you know, even 98% ain’t all that good because you still need that other 2% and you’ll be chasing that, you know forever if you can’t figure out exactly where it is.

So. You know, there are, there is a difference in the kinds of work that it’s doing. So in some ways it was a good test of it because it’s an area where it’s strong. On the other hand, I would also argue it may not be the best because there is very little margin for error in coding because you still have to get to a, a fixed result at the end, from the logic that you’ve put into place.

Whereas in a written piece or a graphics piece, that’s good enough. Yeah. You’re, that crosses the threshold.

Gini Dietrich: Yeah. 80% is good enough. That’s exactly right.

Chip Griffin: And look, I, I mean, I think if we get to 80% good enough on this show, that’s probably, that’s probably sufficient. I mean,

Gini Dietrich: I need to buy one of those sound machines so I can do like clapping and things for you.

Yeah, I need that.

Chip Griffin: There you go. So, I mean, let’s face it, some weeks for us, 50% is probably good enough. So, yep. Before, before we drop down below that 50% mark, though, I think we’ll, we’ll draw this episode to a conclusion. This has been the Agency Leadership Podcast. I’m Chip Griffin.

Gini Dietrich: I’m Gini Dietrich.

Chip Griffin: And it depends.