How the Chief Creative Officer of an Award-Winning Ad Agency Prompts the Perfect Pitch

0:00So I tried launching an agency, did it for about a year. and I would say if ChatGPT was around, it would still be around and be successful because one of the big flaws that I sort of had at that time was I got stuck. I just got stuck in self doubt. And it was almost like I never really got the plane off the ground because I had so much doubt about the plane that was on the runway. We never really took off. I would have doubts about the name. I would have doubts about what it's about. I would have doubts about everything. I was stuck. And I think as a sort of sparring tool, this has been just so great to get not stuck.

0:37Hey, everyone. I'm Jeff Benjamin, Global Chief Creative Officer at Tombris. We're a Knoxville based ad agency. This year is the number one agency of the year. Henrik, you and I have been friends for a while. We worked together for about a year. We've always been curious about each other, about what we do, and certainly about technology. So I think this is going to be a great conversation. looking forward to talking about some stuff like lately i've been noticing how much it validates us you know says great things makes me feel like a genius i think a lot of that is great and has a role in the creative process but maybe it can also hinder the creative process so i'm really excited about that conversation i know jeff because some of the entrepreneurs that work with that's been very successful um like the founder of row and the founder of public.com they used to work for jeff when they work in the creative industry.

1:27And then at one point they were like, you have to talk to Jeff. He's super cool and creative. And he has been our agency of record for Bark working on a lot of our big launches. And so if anybody's ever seen the Bark Air video, which is hilarious, then it was Jeff and his team that made that. Jeff is one of the creators I know who have big teams and a lot of creative folks around him, but very often kind of like go like, hey, I was chatting with ChatGPT or Claude about this specific thing. And I thought it might be an interesting kind of pal out of like, how does people on the front line actually use this on a day-to-day?

2:06Let's go. So maybe I'll just take from Henrik and just say, when do you decide to go to the team versus go to your ChatGP buddies? Like what's your shortcut heuristic for, do I bring this to a human or do I bring this to AI? Oh, that's an interesting question. and it begs a conversation about my therapist would say this is at this point we have to talk about trust because my instinct actually is every time go to ai first and i'll tell you something like i think maybe part of it comes from you know no one's perfect i've got flaws you know one of the things i'm very aware of is sometimes i struggle not struggle but like what do you ask the team?

2:53How do you ask the team? What are you asking them to do? And instinctually, sometimes what I have in my head is maybe not the right questions that I'm asking. And so in some ways, there's a value in just going to AI first, like a chat GPT or whatever, to figure out what exactly am I going to ask my team to do or pursue or which direction to focus them in on. And so in some ways, like it's been helpful for me to sort of like spar a little bit with the ai to to in my head get the the conversation straight how much is this because it's still agreeable that it always goes like that's an incredible idea jeff i saw i was on instagram and there was like a meme or something and they were saying that like people are complaining that chat gpt is glazing too much So I guess glazing is like a Gen Z term for like great idea or like, wow, or like, you know, and I was like, wait, I don't want them to change that.

3:53I kind of like this aspect of it. I do also 100 % think that specifically for marketing, advertising, there are a lot of great answers to any problem and they're all correct. The thing that's really important is that you just sort of pick one. Pick one of these angles and give yourself enough runway to throw yourself at it at 110%. I think you have to sort of be aware that like maybe it is a little self-congratulatory with some of your insights or first thoughts. But I like that it gives you confidence to pursue a direction and throw yourself at it 110 % and see what comes out of that. You know, I really resonate with that, Jeff.

4:35I've noticed recently you talk about sparring partner and I would love to hear if you have a kind of a tangible example. I'll give a couple just to spark the conversation. I have found in a couple of negotiations where it's very ambiguous about what the right course of action is. I've noticed a tendency in myself when it's ambiguous to perseverate, you know, and to mull it over rather than take action. And the, you could call it the agreeableness for lack of a better word, or the enthusiasm of a AI model actually gives me the boost I need to take action. And to your point, I think acting is actually more important, you know, even than which action to take sometimes.

5:14And I've noticed many times it's like, I mean, imagine if I had Henrik always available to me and we're sparring and we're kind of kicking around ideas. And he's like, dude, you should totally do that. I think I would take a lot more action if I had somebody like Henrik, just always encouraging me. And in a way, it's kind of like that. And TBD perhaps, whether it encourages us in the right direction, but it does feel great to kind of get the kick in the, you know, self-esteem to fire off that email or move ahead with that plan. What have been some of your kind of sparring? What are the contents of your sparring and what has it produced or encouraged that might have otherwise gone unattended?

5:53I mean, recently we were pitching Arby's for their social work. And, you know, there were a few different directions that we sort of had hunches on. And I think I was sort of maybe internally struggling with some of it a little bit. And then I started just sort of jamming out a little bit with chat. And you get past sort of the first thoughts, the first ideas, and you're like, wait a second, there's something here. And then I'm trying to avoid details because none of the work's out there. But you start to quickly get confidence in the idea. Then when you have confidence in the idea, you start coming up with other ideas and other branches and ways it can go.

6:34and like I it was just sort of awesome the other thing I will say is that like in my head specifically for me I have the opposite going on in my head being a creative is especially like it's lonely and you know you can have a lot of self-doubts and things like that and so that's in my head and so to have something that's the opposite of that has been awesome for me personally as a creative. When Henrik and I met, I had just sort of shut down my own agency. So I tried launching an agency, did it for about a year. And I would say if ChatGPT was around, it would still be around and be successful because, you know, one of the big flaws that I sort of had at that time was I got stuck.

7:19I just got stuck in self-doubt. And it was almost like I never really got the plane off the ground because I had so much doubt about the plane that was on the runway. We never really took off. I would have doubts about the name. I would have doubts about what it's about. I would have doubts about everything. I was stuck. And I think as a sort of sparring tool, this has been just so great to get not stuck. I think there's another element of that non-stuckiness, which I enjoy personally. I come up with half-assed ideas often. and I've been lucky in my career to be around people who can kind of Henrik whisper so like I'll say something and then they'll say oh you know you mean and then they'll say something much better and I'll go like yeah that's exactly what I meant um but I think what I can do now is I can take a lot of these half-assed ideas and I can do two things with them I can both say hey play back to me what you just heard I said and then when I read it it's easier for me to actually compute if that's what I meant.

8:20And the second is I can then say, imagine you're Jeremy Audley, you know, professor at Stanford. I'm about to encourage him to do this specific thing. Kind of what are some of my blind spots? And, you know, often they will say obvious stuff, but, you know, like often it will kind of go like, you know, to your point, Jeff, it'll come up with another angle that I hadn't really thought about. And then I'll kind of like do a few iterations of that. And so I kind of like this idea of, you know, seeing it as a, yeah and I guess uh as an unstuckiness uh component yeah yeah I mean you said something else too that I think is interesting someone had given somebody else some advice on working with me you were like you know what it's actually forget about what Jeff is saying maybe think about why he's saying it and I I do think that chat GPT has an ability like when you're like throwing things out there to extract like the important thought in your thought that you're maybe not the best at communicating sometimes.

9:17And then it repeats it back to you and you're like, the idea. Extract the essence. Yeah, extract the essence. Yeah. I've had that experience too. It's like, that's what I wanted to say, but I couldn't see it until you said it. And that's why it's been a really helpful tool to go to first before I talk to human beings. And Jeremy, just so you know, this is definitely the creative force between either we're one, the best agency of the U.S. for multiple years in a row, right? Well, this year we were agency of the year, number one agency of the year by AdAge, and last year we were independent agency of the year.

9:53And it's been like a little bit of a five-year renaissance. It's a five-year renaissance. It's like an 80-year-old agency based in Knoxville, Tennessee, which is the last place you'd ever think of. Yeehaw, that's good. And so we've had a little bit of a, yeah. And how much do you think, I mean, like, obviously, I've met a lot of people from your company, and they're all incredible people. But how much do you think that that kind of like openness to embrace technology is part of that story? Or is it kind of like the other way around? Like you are so successful because that you kind of have the openness to explore a lot of different things.

10:28Yeah. I think we're pretty curious people. I think being based in Knoxville, Tennessee, you don't get to see what everybody else is doing. So you get to do things your own way. Like you're sort of tucked in the mountains of Knoxville and you can be weird. it is a sort of like if you come to the agency there's a lot of technology which is a little bit bonkers like there's one room that looks like cape canaveral and you're seeing all the searches that are happening in real time in that moment and there are creatives in there who are getting inspired by that so in many ways we embrace data pretty quickly at the same time i think with ai especially with creatives i would say like maybe in the last couple months it's changed but before that I would get the side eye a little bit.

11:10I would go to ChatGPT to do something or AI and stuff like that. You don't feel like the animosity in the creative team around this? I felt it a little bit up till maybe about a month ago. Did anything change? What do you attribute the difference to? I think people started using it for themselves or seeing me throw off things and they're like, wait a second, that is a good idea. Yeah. Could we talk for a second about, I want to go kind of nerdy for a second and talk about cognitive bias because of something you said earlier about giving you confidence. So, you know, at the D school, we use a classic improv technique called yes.

11:52And everybody's familiar with yes. And right. It's like a classic brainstorming technique. And the thing that's interesting about yes. And is it serves two purposes, right? So if we're brainstorming, we all say yes. And right. One purpose it serves is it puts me, the speaker in the posture of, uh, now I got to build by saying yes. And I can't be a doofus and say something irrelevant. So it kind of keeps me focused, you know, and on track and in a building contributing posture. But the other thing it does is it helps you feel safe when you know anything you say, I'm going to respond with yes. And you have a sense of carefreeness and a willingness to express.

12:26And that's what you're getting at with Chad GPT is it's building my confidence. I would say just like humans who are yes anding build your confidence because people, we start feeling invincible. If we start yes anding, we're just like high fiving and chest bumping and having a great time, right? So that's the benefit. The flip side or what I wonder about is what happens in terms of anchoring? So if what we usually have done in the context of a team and the yes and and the confidence building and exploring, if it now happens with AI and then I come to my team, my confidence in my idea is already built perhaps and the team may not have been a part of the exploratory process.

13:04I mean, this is a concept called anchoring, right? Where you become kind of, you start to defend your ideas because you think they're good. What's the impact on your interaction with your team now that ShadGPT is building your confidence individually? Well, I mean, hopefully I've picked a team that's going to agree with me all the time. I think that you, I mean, part of like where I am right now is that I've got to have like some sort of instinct for which way we should go. But I think that when I'm going in there and I'm saying, hey, this is the manifesto or this is the idea, I'm open to the thought that it's not.

13:39And they may have a different point of view and I can have a constructive conversation because we're not actually sort of willy-nilly floating around. we actually have something like a piece of clay that we can mold here now there is sort of a minimal viable product or whatever right there's a thing to react to there's a thing to react to which i think is different when you're just sort of like throwing a random idea out in a brainstorm and it has no gravity you know it has no weight to it i really think it's that's such an interesting point lately i've been thinking a lot about what are the different outcomes of ai and i got obsessed with this kind of meta thought that most people gravitate to concepts that has been articulated clearly and labeled so there's two concepts that we talk a lot about in the future ai it's basically universal basic income and it's some permutation of what i think was called like the paperclip kind of thesis which is basically ai will turn us all into paperclips but these are very known concepts.

14:42And because there are known concepts that are label and are pretty clear defined, I feel everybody just graduate automatically to put the future into those buckets. And what I think is interesting in what you're saying is that what I think AI actually can help sometimes is that it can take a lot of different thought and then it can help create almost this kind of conceptual bucket. So when you come into a meeting and you say, hey, what do you think about name of kind of cool you know like concept with name of three lines of kind of like uh premise for this concept i think it's much easier for people to take all their random thoughts and almost kind of put it in that bucket does that make sense yeah and i think honestly for us creatives it's also kind of part of our job to create buckets that have gravity so that other people kind of can get excited about contributing to them.

15:34Because all these very abstract thoughts are just difficult to have in your mind. And it's difficult to have a good conversation about because kind of abstract conversations just end up having abstract outcomes, right? Well, and ironically, or not, I don't know if it's ironic or not, but I don't know what that word means really. It's just an interesting word to use. But the capacity of an AI to insta bucket something i.e you know here's a bunch of my loose thoughts create a four letter framework that's a that's a helpful mnemonic yeah right and all of a sudden it like feels more legit because ai oh it's the poll acronym p you know pretend you were listening you understand nobody cares anyway l listen like you're listening to your fate right you can just make it up but then people go, oh yeah, the pull framework.

16:26We use that all the time. It's pull, not push. Even before AI and stuff, what we would do is, okay, you're working on a project. Teams are coming in with stuff. You've got literally hundreds of ideas, some smaller, some bigger. You put them all on a wall. Now you start to arrange them to try and group things together because what you're in search of is the big idea you know and you're it's a little bit like homeland carrie matheson yes we're trying to sort of make sense of all of this and it's like okay this is the idea or this is the idea there was a guy famous aurea murkin he was famous for like this guy wouldn't show up for a couple weeks and then he would show two weeks into this thing he'd go to the wall and he'd be like that's the idea my camera's a little bit off uh But AI can do that.

17:19And you can say, you know what? Give me another one. Rearrange these again. Yeah, yeah, exactly. Do you think it can do that picking? And I say this because I talked to somebody who works for a big luxury brand. And he was kind of like said, one of the issues that we find is that we're all about taste. And taste is what we sell. And AI can't do taste. I think it's a superficial argument. I agree. I agree taste is important. but taste is a function of experience, which means it is a trainable thing. And insofar as language models are trainable, you can train them. I think that's a little bit of a cop-out respectfully.

17:56I don't know who this person is, but I would respectfully suggest, or I would ask the question, what if you could imbue a model with your taste? What could you do then? And anytime we're like holding onto the thing that we say, this is the unique thing we do. I think that we're in a dangerous position. I think when you're more specific about the audience, you know its taste gets better I think its taste is better than many other human beings now but here's what I do think Henrik is that I'm actually just asking it to create some arrangements and I will use my taste you know to decide if that's right I'll ask it to rearrange it if I need to but ultimately I'm sort of making the end judgment and in that sense you know like I love this word because I've been trying to like figure out like so our Our platform for the agency is data plus creativity for business results.

18:47I was like, how do I make this thing exciting? Well, it was there before I got there. How do I make it exciting for creatives? So before AI, I was playing around with this word, and I feel this way even more now. I love this word bionic, where human beings are augmented by technology. And that's what I think of that relationship. When it's coming up with these different variations, ultimately, I still have to put my humanity and my instinct and judgment and taste, I have to apply that to this. There's a weird tendency, I think a desire for it to be so magical that we don't need to do any work.

19:23I've observed that. Even, I mean, Henrik knows, we conducted some research last year, myself and a research colleague out of Harvard. We conducted this research. What we found was the teams that exerted the least effort felt the best and had the worst outcomes. And the teams that exerted the most effort felt the worst and had the best outcomes. Yeah. And so, and there's all, it's like AI, it's just got to make it easy, right? And what we don't want is we don't want to expend the calories and granted, you know, going to the board and finding the great idea is an expensive task, right? And if you say, now rearrange them and find it again, it's like for a human, they're like, dude, I got to go get another Red Bull.

20:01I need to take another walk, right? But for an AI, it goes, sure, happy to rearrange, right? But then you still, the questioner, as the human, will you still say now I'm going to exert the effort to evaluate the rearrangement? And that's where a lot of folks underperform because eventually you're just like, you know, whatever it says, right? Because it's cheaper from a calorie burning perspective. Yeah. I sat on the board of Biaga Angles group for many years and Biaga has this kind of cool, he's a big architect for people who don't know. And he has this kind of expression, he talks about people doing homework, which is basically when he comes into his studios and people have kind of answered whatever the questions that they were given, but it's kind of like just something they did fast because they wanted to go and do something else.

20:47And so based on that, I now often kind of like my question when people give me something, maybe even before I see it, it's like, would you put this on your portfolio? Like, is this going on your portfolio or not? And they were like, maybe not. It's like, okay, you know, come back when it goes on your portfolio. So I think it's interesting. And I have a question around that. That's a great question. We have actually, I'll just tell a short story from a podcast we did like a few weeks ago where we talked, I think it was the CEO of The Atlantic. And we got into this very interesting kind of conversation, which is basically what are the human skills that you need to practice for you to become better using AI?

21:20And the short thesis is that let's say that you are bionic and you have an Iron Man suit on. And if you can lift 200 pounds as a human with a bionic suit on, you can lift a thousand. But it's an exponential technology. So if you can only lift 50 pounds, then you can only lift, let's say, like 250 pounds. And so it's a multiple, but there's still a base rate that's multiplied. Yeah. So the question, Jeff, is I'm sure you've seen many ideas that come to you that is kind of clear chat DBT, right? Like if you can just scringe your eyes and you go like, OK, this is pretty average. What do you think is some of the unique human traits that we as creatives have to practice in order to get the best out of the bionic suit that AI has given us?

22:08I think asking better questions, more specific, like what is the voice of the question you want? Like one of my little tricks lately is like, how would Don Draper answer this? and part of it is like throwing a little bit of like a crazy variable into the prompt that's going to get like a very specific and different more inspiring type of answer to a problem so i think there's that i think it's also just asking lots of stuff you know one of my things with you is making is thinking right and so how do you turn this relationship you have with ai not in it solving the problem, but it being part of the making process.

22:50It's just like Word or Photoshop. It's not actually doing it. It's just a tool to help you do it. My visual language is always stone carving that kind of, I start with the empty prompt, which is basically just like a big block of rock. And then all my questions is just me carving stuff away. And sometimes I go like, oh, this will be a pretty woman. And then I accidentally cut off the nose and go, okay, not a woman anymore. Now it's something else. Right. And so it's all about just kind of like following the flow, uh, as you kind of like build it out. Yeah. Yeah. There's something interesting when you say, how would Don Draper answer this, Jeff, which I think is worth calling out, which is there are certain, and I'd be curious if you have other examples of this, but there are certain you could call them what a nerd would call them as metacognitive strategy, which is thinking about my thinking.

23:39And what you realize in that moment is, oh, I need an expert perspective, right? And we talked at length with this amazing dean of Harvard's education school, Stephen Kosselin, you know, a few months back about going upstream of our thinking. And a lot of times the really interesting insights are when we kind of push our thinking upstream from the outcome to a more specific prompt like that. And that I think is a function of experience, right? Knowing to ask the question, how would Don Draper think about this, is valuable. And again, it's expensive because probably like you've got some output that either you can think about, which requires a certain amount of caloric energy, right?

24:19Or you can ask an additional question, which is going to force you to kind of reactivate and re-engage. But it's more expensive, probably. How do you think about when to inject more questions versus when to just stop or when to go to a human? Oh, I'm just a curious person. So I'm just, in some ways I was made for this moment. Because it's tireless and it's willingness to continue to. I read the end of it. It's like asking questions and it'll always give you an enthusiastic response. But when do you go? Now I can go to the team. When I have enough to be like, holy, like, you know, like this, this feels like an idea or this feels like a bunch of ideas or I think we're, I think we're onto something.

25:07There seems to be enough here. You know, Henrik said something before. I don't know if this is going to answer your question. He said something before and it made me think, like, this is not advertising related. I've got an idea for a rom-com. All I have is a title. I wish I had someone for you. That's the title. I think it's a good rom-com. I don't, I'm not sure. You know, so like one of the first things I did with ChatGPT was I said, like, write a screenplay. Like, I don't know the first thing about writing a screenplay or anything like that. I just got a hunch that this title is a good title.

25:38And you're like, oh, my God. Like, it's like, this seems like it could be a movie. And then I'm like, well, write another one. And then it writes another one. It's completely different. But I'm like, this is a pretty good movie, too. Like, you do that enough times. You're like, wait a second. I got something here. I had a theory, a hunch that I wish I had someone for you was good. but like I'm seeing enough here now that like there's fertile territory. What I think of as a big idea is an idea that can give you lots of ideas. I should be able to go to the media team. I should be able to go to the programmer and they should have ideas when I tell them this idea.

26:11Like for me, a big idea is something that's got a lot of width to it. And so because of that version of what I think a big idea is, when I'm seeing a lot of stuff coming out of this computer, that seems pretty good. You know, maybe it's not like the can-winning idea or the idea that's going to break the internet. But when I'm seeing enough, that seems like, wait, this is a place to go fishing. I also think, you know, I don't know, when you were asking that question, Jeremy, I was kind of thinking, I wonder when I kind of stop prompting and start like talking to humans. I think it's when kind of like the excitement for what I got is bubbling over.

26:47Yeah. And I need to share it with somebody. Yeah. And I don't necessarily, I don't need to share anything with chat GPT, right? You know, I'll just always be there waiting. Like, it's not like, I'm so excited to sell you this new idea. You know, chat, hold your GPU. It's not actually a GPU. So anyway, you know, my dad once told me, I asked him once, like, when do you ask somebody to marry you? And he had like a Danish phrase, which was something to his end of like, when there's nothing else that you would rather do. Like when that is the only thing that kind of like obsesses your mind, right? When you can't help yourself, basically.

27:21And so I think, and I think what's interesting with, you know, like this thing about when do you tell a human is that it actually talks back to the point about what is the human trade that we should practice. And there's something very innate, I think, in most humans that when they have something exciting, they want to share it. And that point, that boiling point, the hundred degrees Celsius of ideation, when it suddenly goes from being water to being labor, then I think it's kind of like a fascinating point. Yeah. I mean, before any of this, like I'd be at work, I was like a 22 year old graphic designer and I'd be working all night or like a few hours on something and I'd be like, it turned a person next to me.

28:06Oh, do you want to see something cool? Can I show you something cool? I think the greatest conference room in any creative company is the hallway. You know, you're bumping into someone like, what are you working on? You want to see something cool? Like, do you think this is neat? I'm like, to your point, Henrik, I think there's a natural moment where you're like, oh my God, like, look what it did. You know, I think we're onto something. I have to share this with someone. So here's an interesting question. I'm just, cause I'm thinking about what does that mean? Is the urge to share really because I need to take action, right?