Prompting for Originality: How Devin McNulty Is AI-Enabling His Workshop Business

0:00Hi, my name is Devin McNulty, and I own a company called Funmentum Labs. And we've been bringing fun into work for 15 years at really, really big companies and using it to get results. And the last two years, I've been taking my workshop fun method and automating it into a software called Funware, which allows teams to tap into some of these insights and run these mini workshops on their own. and I've used a ton of knowledge I've learned from beyond the prompt to help me figure out how to do it. Devin, Professor McNulty, okay. I want to dive in because you've done something. So I've been using your tools for a while and I am quite delighted at how much fun you bring to the workplace.

0:50And I wonder whether you could just talk for a second about fun as a story of the points. Yes, absolutely. For me, my obsession is not fun that's kind of extra or frivolous or on the side, but fun to try to get some of the hardest work done, such as like landing our strategic vision and then making that actually matter in the week to week work of a company or figuring out what our big 360 marketing campaign is going to be that we're going to spend millions of dollars on. And using these fun techniques to try to get groups exploring new avenues that they might not normally, and then trying to align faster, sometimes cutting out months out of a process, just by getting people into a different headspace.

1:37There's so many questions I want to ask you. what objections get raised when you lead with fun? I can imagine if I think about the hugest entertainment company or the hugest financial services company, you lead with fun and there's probably some resistance. Can you talk about the resistance? Why fun? What resistance? How do you overcome it? Yeah, yeah. Well, one of the things that I've been lucky enough is just to have so much social proof at this point that I can just point to like, I've done this, this, this, this, this, this, and this and talk to this person. So that's an easy way to shortcut that.

2:11But your main question is, one, I don't have time for it. Or we did something that was fun and not much came out of it. Or like, I got to, you know, I got to go to this meeting, I got to get this thing done. And I'm a serious person. But what I oftentimes love, okay, so I was doing this big presentation or this big interactive training workshop, a three hour thing for like 250 biotech executives. is. And the thing that the main person told me right before I went on stage is he said, there's going to be a lot of really skeptical people of you out there. And I was like, thanks. Great. I appreciate you saying that to me right before I go on stage.

2:50But then, you know, so it's like, I always know, it's almost like the body language starts like this. And for listeners, it's like, I'm just sitting back and folding my arms and furring my brow. And then once you just get people, I just ask people, just give me a little bit of your energy and just try to start. And if you're still skeptical at the end, come up and talk to me. But what ends up happening is like the body language slowly changes, people start getting into it. And then by the end, I get mobbed with people just asking me like, how do we work more like this? And it can be a little tricky because there's very certain techniques like around like creativity and psychological safety and specific mechanisms to try to actually use the hard parts of work and to turn it into something that feels a little fun.

3:37I'm not sure if I answered all your questions, Jeremy, but that was me. You're getting there. I mean, I think this is the heart of so much of what we believe, right, in ideation and creativity is dispelling this notion that if it's fun, it's probably not work. Or the reason it's work is because it's supposed to not be fun, right? And you just, you lead with fun, which obviously we're deep believers of. What would you say, I love that, you know, kind of furrowed brow, crossed arms. what are your kind of go-to tactics to shift someone's posture set aside they're totally embracing just how do you address the skeptics and what are your first moves so to speak when you see a room full of 250 biotech execs with their arms crossed yes um well one of the first moves is i try to preempt that completely so i would recommend doing cold opens where when people start coming in, you just start pairing them up immediately and giving them something that feels like a fun way to explore something that's real.

4:38So you might say like, all right, I'm just going to pair you up with somebody and I want you to discuss a time like a memory related to getting rejected for a credit card. If the larger exploration is around how do we get more people to sign up for our credit cards? And so there's just this way of it's almost like exploration around like you can go directly at something or you can break the script and try to get people exploring or bringing up memories or especially funny memories around something that will ultimately come back around and serve as fodder for your larger question. Another one, and this will be nothing new for you, Jeremy, but is just ask people to just generate lots and lots and lots and lots of thinking, whether that's with visual thinking.

5:26So just putting lots of things up on the walls or exploring with mind maps and then giving people some space on their own to kind of think of something and then pairing them up with people that are outside of their world and having them give feedback and building something big. It's funny because like in some ways what I'm describing might not necessarily sound fun. But when you do it in a way that moves really fast, like especially for the Zoom ones that we do, you just invariably get people saying, wow, that didn't drag. And I think that's a huge part, especially for some of these executives.

6:07It's like there are meetings all day and it feels like the conversations are circular. And so if you can have a moment that felt like, whoa, like we didn't perseverate or politic or we didn't spend too long on any one thing um it's almost like that's even more fun than having a happy hour where we're all just kind of chit-chatting which i think is what a lot of people think of when they think of fun at work can i ask a question evan um and maybe it's a little bit of a change of tone but what is ai funny um no well not really it's kind of funny right um and so you know we'll of course get into like like my journey into ai um but i found ai can be really like with very certain specific prompting you can create a container where it can be funny but if you ask it to tell jokes that are going to move an important corporate conversation forward no way henrik wasn't that one of steven johnson's kind of observations right When we had Stephen from Google Notebook LM, he talked about how it has yet to be funny.

7:14I think, by the way, as I saw a paper last night just posted by Ethan on LinkedIn. Shout out to an earlier podcast guest, Ethan Malik. But he he referenced how now AIs are creating memes that are judged by humans to be funnier than human memes, which seems like an advance, like breaking news kind of moment. But just for the record, it seems this may be an evolving, you know, part of the jagged frontier, so to speak. Yeah, no, I actually love that because I think that that really is an advance because if you can hit humor, it's almost like higher, higher order thinking of like, oh, like you can you get a certain sense of me.

7:55You get a certain sense of culture. I hadn't seen that thing from Ethan yet, who's obviously amazing. But that is that that's next level. I haven't seen that yet. let's go into the ai journey and maybe maybe very simply we can go backwards to your journey but what role does ai now play i mean folks now kind of have a sense for you're bringing fun as a way to advance far more rapidly than an environment without fun what role does ai play in how you do that like how do you use ai to accelerate your own acceleration process yes so basically um we We created a product called Funware, which is designed for teams of five to seven people in hybrid situations to be able to run workshops on their own, brainstorming workshops, retrospectives, lots of other kind of workshops key in the project management lifecycle.

8:49And so what we're using AI for is these little moments where, because oftentimes if you think of AI, it's like a, it can be mainly a solo pursuit. And I use it, of course, all day, every day, but it's mainly me going back and forth with it. And it's these like long conversations and long prompts. But with Funware, we like to have, essentially, it's like an AI button and it does one thing really well. So, for example, when I'm starting a workshop and I have a prompt such as, OK, we need ideas to get more listeners for our AI podcast beyond the prompts, just to choose a random example. You know, we have a button that's like, ask the AI robot to rephrase your challenge.

9:35And so then the AI robot will give some other rephrasing challenges such as I just did this right now. You know, it says, hey, here's a different way you could say this. how do we utilize social media strategies to increase our listeners? Or how do we create partnerships to expand listenership? And you might say... There's quite a lot of people who have asked to have like a semi-nude calendar of Jeremy, which we're kind of working on for the fall. I don't know if that will work. That's what you've been doing in mid-journey, isn't it, Henrik? To the point where my wife is getting really, really concerned.

10:08There wasn't jealousy at the beginning of our Beyond the Prompt journey. There was no jealousy. There's a fair amount of jealousy now. and so for that one that one might be a little bit straighter it's like where's the fun in that and then afterwards when you start doing like a bad idea brainstorm the AI gets in there with you and I've done like many many many prompts to get it to the point where it just automatically comes up with bad ideas that also have like a sliver of insights and are kind of like slightly fun and so they're designed to push the group's thinking so you've got four five six humans in there that are all coming up with bad ideas.

10:46And you know that it oftentimes takes a force of personality facilitator like me to get people into that headspace of like, no, I know you're at work, but I want you to actually come up with bad ideas. And so the question of could I try to create that environment where I'm not there and it's just a group on their own doing it while I'm trying to make it so the AI kind of greases the wheels a little bit, starts coming up with bad ideas that are fun or with a different exercise where you choose a celebrity like Beyonce or Garth Brooks. And now they're a version of them in the AI is coming up with ideas with you for how to get more listeners for beyond the prompt or, you know, how to improve product features or whatever you're thinking about.

11:30And I've seen again and again with groups how the AI injects that little bit of levity, but is not designed just to be frivolous. It's designed to get the group actually opening up their thinking and pushing in different directions. Okay. So I just got to interrupt you here, Devin, because I've used your tool. I've used it in workshops and sessions, design thinking stuff. And what I would say is the AI, you said it well, what historically would require someone who's a facilitator like me, dot, dot, dot, right? That's kind of, that's a phrase you used in that description. I think one thing that's interesting is AI as the person who's responsible for the container or responsible for the interaction, which is a very different kind of framing for AI than a lot of, like people, you know, might treat AI like a thought partner, right?

12:18But actually handing over the keys of facilitation. And the other thing I just want to call out here is most folks, you know, if they say how to bring AI into a brainstorm, it's like, okay, now we're each of us individually also having like a chat GPT window open, right? And what you've done with Flimware, which I think is super cool, is you have said, no, this is an environment where the group comes and AI is actually going to facilitate you in a group setting in this environment. Can you talk for a second about this role of displacing yourself, for lack of a better word? Because basically, in a sense, you just hired AI out to do your job, right?

12:54To use a phrase. Yeah, yeah. Well, in some ways, it's like I would hope to be able to get to the point where it can much more replace me. Like right now, I'm primarily kind of coding these exercise flows and then I'm injecting AI in little specific moments like, hey, come up with bad ideas with me. Help me rephrase this challenge prompt. Come up with good ideas. At the end, when everybody's put things in, synthesize them into some final insights for folks. But eventually, I would love to have the AI live. Like, okay, so Jeremy, you know how as a facilitator, sometimes you have your plan in place and then you get in with a group and you say, you know what?

13:38I'm changing on the fly. I'm actually going to do this exercise instead. Like, I would love to get the AI to that point. We're not there yet. Where the AI is actually gauging folks' interactions or maybe even lack of interaction. And then on the fly says, you had asked me, you know, a person who handed over facilitation to me, you had asked me to accomplish this. I'm now sensing that the group actually needs this first, perhaps in order to get to that goal. Yes, exactly. Exactly right. And so I'd actually love to just give a little bit of my story to get into AI. Please. And just give people a sense.

14:11So, you know, I've been running these workshops for a long time. And before the pandemic, it was always in person, always in person. And it would be this big thing. We get together 50 people in the cool photography studio in San Francisco. and we create this four hour peak experience. And then the pandemic happened and we said, oh no, what are we going to do? We adjusted and adapted using Zoom and Google Slides and created a really powerful virtual experience. So then we started asking, what if we could automate this? And we had a former client of ours at Atlassian who came in and said, you know what?

14:48I love this so much. I'm going to come on as a technical co-founder for you and I'm going to build this with you because I, as your client in the past, loved this. And then from there, that's all of a sudden when AI came on the scene, for me at least, I think it was ChatGPT 3.0, and we started saying, oh, we need to be able to figure out how to integrate AI into this. You were asking the question, how do we automate this before AI? Yes. That's cool. Before, I mean, AI was around, but I was in a comedy group for a long time, And, you know, maybe six, seven years ago, a forward looking member like took all of our sketch comedy scripts and put it into this like early, early version.

15:29And it created all these garbled sketches. And it was a it was just something we laughed at. We were like, this is so bad. But we're not laughing as much anymore, at least about this part. And so from there, I actually set me off on a journey. I said, I need to learn how to do all of this complicated creativity and AI stuff. And I started watching tons and tons and tons of YouTube videos about it. And that eventually led me to you, Jeremy. And I deemed you the best expert on creativity and AI. And I got obsessed with Beyond the Prompt. I've listened to every episode. And it's taught me so much.

16:10I've learned so much from this podcast. Yeah. And then I'd love to actually share some of the complicated prompting things for the nerds that I'm doing in the back end of this software. That's all right. Yeah, maybe walk us through a couple of kind of key moments in a workshop because now listeners are hopefully kind of getting the sense. What's special about this conversation is Devin is a listener just like you, so to speak, right? Who has been wondering, how do I apply these tools to this core workflow that I've got? And leveraging some of the resources that we put into the world and that many others like Ethan and Connor, lots of amazing people have put in the world.

16:46Devin's actually doing it, right? And so you can think about Devin as if it's a mountaineering journey, he's just a little bit farther up the mountain than you are, so to speak. So maybe, Devin, if you wouldn't mind, could you take us through one or two kind of key facilitation moments? What would you do as a human? And then what is the process by which you've been getting or been learning how to get AI to do it? Yes, yes. So for the moment when somebody enters a challenge, such as how do we get more listeners for our AI podcast? Or how can we improve our product based on user feedback? Or, you know, what would be great 360 marketing campaigns for a AAA video game?

17:27Whatever it is, there's this button that says, ask a robot to rephrase your challenge. And the prompt itself is probably about like seven paragraphs long of what's happening in the background. So to super high level it, what I'm asking it to do is I'm asking it to take what the user inputted to figure out kind of what industry they're in, to try to gain as much context as it can imagine about it. Oftentimes people are putting in seven words, so not much. And then to create a bunch of fodder. This is all happening in the background. The user never sees any of this. to create a bunch of fodder of how you might make it, the component parts of it.

18:07So if it's get more listeners, some component parts might be like social media strategies or partnerships or AI communities. Maybe component parts isn't the right word. It's more like slightly more specific avenues for how to explore. Because Jeremy, as you and I know, oftentimes people come in with ideation challenges that are a little too generic. And so if you can get people to something that's a little more specific, you can oftentimes get better ideas out of it. So I'm trying to use the AI to help make sure to basically push groups into slightly more specific ideation challenges. And so in the background, it's taking what they put.

18:48It's coming up with lots and lots and lots of fodder. And then it's deciding on three final like replacement challenges for the brainstorm. And then I ask it to evaluate those challenges on like four different criteria and then plus it up based on that. And then at the very end, to just give the people in the brainstorm the final three rephrased challenges. And it's even more complicated than that, but that's like the simplest version. Does that make sense? Yeah. What are tricks that you've discovered to atomize the process that you are already doing so that you understand where you should apply AI and where you shouldn't?

19:38Yeah. Yeah, well, I think a lot of this, and this is that interesting intersection between, you know, 10, 15 plus years of just knowledge of how to run these sessions and hundreds and hundreds of them, like that very human knowledge and my intuition of here's what I, here's what I, here's that top of the mountain that I want to get to. and then lots of trial and error with AI. And it's interesting just listening to podcasts like this and watching tons of other YouTube videos and sort of just getting inspired by what other people are doing and asking myself the question, how do I get there myself is one way.

20:14And then of course, like the most fundamental insight, which is ask AI how to do it. And so it's that constant iteration with the AI where I'm saying, here's the existing prompt. Here's the input that users put into this, the fake users that I'm constantly testing with. Here's what we got back. Here's why I like this. Here's why I don't like this. What are we going to do to make this stronger? And then the AI is working on it with me to get that prompt sort of perfect. Here's a very, very nerdy question. I'll give you a little bit of background. So over the weekend, I was working on, I got ChatGPT to write me a Google Apps script that basically check all the emails I get that has a newsletter markings.

21:02And then based on my kind of like things that I'm interested in, it basically summarized all the newsletters I've got over the last 12 hours so that when I wake up in the morning, I can basically just scan if there's something interesting. And I don't really know how to code. So, you know, like I just use ChatGPT to do that. And every time, you know, I just bytecode myself into that. What I noticed a few times was that it kind of like wasn't really giving me the right results. And so at one point I needed it to kind of completely change the strategy on how it was approaching the problem. And it was actually, I found it to be, if I asked it to change a strategy, it was almost like that was like a meta concept that it kind of struck a little bit with.

21:44have you found ways of saying hey i'm going to use ai to reframe for example like the problem people come in with or kind of give them really dumb questions have you found a way of kind of exploring what are the different methods on which i can prompt it in order to do that does the question make sense oh it makes so much sense um this is where i've actually found some of the limitations in the models too where i say okay give me like five different strategies for how to do this and um sometimes it has interesting ones also i have to say i did some of this work maybe nine months ago and so at the the rate that this stuff is changing um i bet you now if i came back to it it would have been a much better partner so it's almost like i have to put myself into the month of like oh i think this was last june and so um but i think now actually yes like if but it's interesting it's almost like you don't want to overload the ai with too many things at once.

22:43So you might ask something like, hey, I'm trying to figure out this way to get a more interesting challenge reframe. Can you give me three different strategies? Here's everything I've tried so far and just copy paste like pages of contact. Can you help me figure out three alternate like ways to get at this? And then I would also just nowadays I take everything from Jeff Woods by everything. I mean that ask me first the interview questions. You want to talk about a very specific thing I got at this podcast. And then, of course, immediately read that AI-driven leader, such an amazing book. Devin's referencing a previous podcast with Jeff Woods, where he talks about leveraging AI as a strategic thought partner.

23:21It's a great episode to listen to. Then he just referenced Jeff's book, The AI-driven Leader, which is a great book. But Devin, I just wanted to do sidebar on the sidebar. You're kind of doing a shout out. I just wanted to make it explicit. You're saying you listened to the Jeff Woods episode. And then in the context of Henrik's question of how to change strategy, how would you leverage your insights from the Jeff episode to Henrik? So what I would do is I would say, I'm trying to change how I prompt AI in the background. Here's what's working. Here's what's not. Before you give me three alternate ways to prompt this, ask me three questions one at a time to figure out what it is I really want to accomplish.

24:00And this is straight from that episode in that book. I mean, like, and I definitely use that where I still think sometimes, and as we were talking about, is sometimes that the extremities of the outlier of the ideation, I find needs kind of like outlying prompting. And so, for example, you were saying, give me really bad ideas, which I think is a very good prompt because then it doesn't give you the generic kind of like suggestion back. if I tend to use, please give me something that is completely illegal. And not because I necessarily want to do something illegal, but just because it suddenly prompts, like it goes a different place in its model to try to kind of like reply to that.

24:40You're prompting for variation. Exactly, right. And I think increasingly as we're trying to be original, what we're trying to do is to cast a net or give the AI the ability to kind of go a place in its AI brain that isn't the standard answer. It is kind of like the outlier answer because then we can discard most of the answers, but suddenly we'll see something that we haven't thought about it. And because AI is not tethered to this world, like sometimes you can go quite places. Like what would a microorganism think about this specific thing? And it'll obviously answer something back and then that might kind of prompt you to prompt it back something else.

25:19And so it's just curious is as you're trying to kind of become that kind of great facilitator, which obviously what facilitators obviously do is that they kind of like take the conversation, all these kind of like different entry points, and then suddenly something kind of new, something new kind of materializes. Well, Henrik, one thing that you're reminding me of too, is that the, a lot of times we've actually got to think about our thinking in a way, you know, and I think we discussed this on the episode with Kian and Dave McCraney, but human cognitive bias is a lot of times the biggest limitation in a collaboration with AI, right?

25:54And specifically the bias I think that's at play here is you just want the right answer. Most people, when they come to a collaboration with AI, aren't actually looking for volume or diversity. It's just like, give me the answer. That's a human cognitive bias. And if you become aware that your natural human cognitive bias is to be as lazy as possible and to do as little work as possible, Then you can say, actually, don't be lazy. And I'm prepared for a back and forth, you know, that's five or 10 rounds long. Right. But that, you know, it's similar to Devin's prompt of give me bad ideas. It requires an awareness of the human cognitive bias to say, this is going to help me.

26:31It's going to help me show up differently as a human to have AI work this way. It's kind of similar to what Russ Summers told us in his podcast, you know, when he names all of his GPTs and his GP team, he names them all and gives them all personality. So it doesn't change what they do. but it changes how I show up as a human. Therefore, it changes what they do, right? Amazing. Another of the episodes that totally blew my mind and changed my life. So to bring it back to fun, we have an exercise where for your challenge, you can have like a celebrity that's now the AI coming up with ideas with you.

27:05And Jeremy, what you're saying is exactly right. It's hard to think. It's hard to try to come up with creative ideas and our brain is a cognitive miser. so we want to immediately jump to the first thing that like feels like it makes sense but when you have an ai partner in there that is just coming up with like off the wall or like different ways of thinking about your challenge um sometimes you'll read it and you're like that's actually really fun or like that's interesting or that's inspiring or that gives me a little bit of a something to jump off of and so that's almost a way to rather than you know having a stick and telling people like, you must think of more ideas.

27:41It's almost a way to use positive psychology to pull some of that thinking out of you. You are just going to settle and go for like, essentially the thing you did last year. But now that the AI is there, and it's been programmed in the super complicated way to give you like, push it this way, push the conversation this way, push it this way. I've seen groups take something the AI says, say, that's not quite right, but that's actually really interesting and then that leads them to a new place yeah one thing that uh now i'm kind of a little bit kind of uh stuck in this idea of strategies for prompting ai to give us original kind of like uh information back one thing that we've noticed we have this business called autos where we help people build ai businesses um and one of the elements we do is we create like the logo for people.