Phillip Maggs maps the future of design + 20 lessons from our first 20 episodes

[00:00:01] What is the future of design? After seeing creative technologist Phillip Maggs present at Figma's Config Conference, we just had to interview him. We discussed industry disrupting design automations and how they empower human decision making.

[00:00:16] In the last two years, I can't even write down the amount of use cases I can think of for large language models and action models. There's not enough hours in the day to think of the use cases. There's so many. I'm not trying to convince people that there is a use case. Now I'm trying to convince people that we can do it.

[00:00:34] In episode 20, we answer the questions that have been worrying designers and product teams alike. Which types of design will be automated? Design systems and brand guidelines as LLMs. How brands will use Gen AI to augment specific workflows? Which skills, mindsets, expertise will protect your job? The best way to leverage Gen AI as a force of good? And the importance of becoming a creative technologist.

[00:01:00] But I still think it would do a worse job than a human would because a human has way for emotional context to why something is resonant. How do we visually brief a human? So for example, like the main part that I'm working on is how do you go from brief to first creative draft and first creative review in order of magnitude time?

[00:01:17] Phillip Maggs leads generative AI excellence, empowering Superside and its customers in harnessing the potential of AI. Built, tested and developed, Superside's AI service offering with 500 plus projects and counting.

[00:01:30] Many of which are to augment the creative teams of the biggest brands in the world. He's worked with Netflix, Google, IBM and Rolls Royce. Plus he's an award-winning strategic consultant and special effects company co-founder.

[00:01:42] If you've got a thing that has to last for three hours in an email marketing campaign and you've made it in three minutes, there's a really good relationship between what the material is intended for and how long it takes you to make it. And the same with brand as well.

[00:01:57] Thank you for listening to the Design of AI podcast. We interview AI leaders and discuss the latest innovations. We help teams learn how to leverage AI to reshape their industries. If you like learning about AI and how to grow your career, make sure to follow us on your favorite podcasting app. It really helps us if you leave reviews. To get more AI and career resources, subscribe on Substack at designofai.substack.com.

[00:02:20] Hello, Philip. I saw your talk at the Figma config. You presented a future of design that really was shocking, but it makes so much sense.

[00:02:32] So today's conversation is really aimed to be a framework to help designers imagine design through your eyes. What you see coming in that future will really be trying to talk about what designers can do to prepare themselves.

[00:02:46] Perhaps what they might be worried about because the reality is AI is creating shockwaves and those shockwaves are only going to keep growing as much as some people say otherwise. This is real. This is happening. Everybody's building these tools. Every week there's new releases. Adobe yesterday launched some crazy new tools.

[00:03:04] You're telling me about Figma launching some new tools. So I'd love to know where you sit at in the philosophical realm of AI right now.

[00:03:13] Thank you for having me. First off, I would say bullish is where my general philosophy is. And that is backed by quite a lot of hands on experience over the last two years. So I've yet to see anything that has not either expanded out opportunity from trying out these new tools in new ways, specifically in professional pipelines.

[00:03:36] I'm not talking about like abstract areas of AI and writing a book or AI and making art, talking about quite specific areas of a brand needs these things tomorrow.

[00:03:52] And they have these real constraints and they have to use this software and it has to work in this bit. And I have to be able to edit that bit, which is just the reality of general professional design work is that you can do it.

[00:04:06] You don't get to use everything you don't get to use everything you don't get to use everything you want to use all the time. You have to work within constraints. I've yet to see anything that hasn't expanded out on possibility from doing it hands on.

[00:04:16] So wherever those edges are, we haven't found them yet. And that kind of leads to a quite risk aggressive, maybe, but quite bullish stance on where the future will go.

[00:04:30] And to put that in detail, your team at Superside is building tools that are getting deployed and large orgs right now to rely on AI and revolutionize workflows, right?

[00:04:42] So, yeah, I mean, yes and no. So Superside by its nature is a services company, but we are a services company backed by technology and Superside wasn't founded by me.

[00:04:54] Our founders, nine years ago, have gone through multiple iterations of what this company is, mainly focused around how do you distribute a full-time workforce around the world to serve big companies, create opportunity for people around the world and serve these amazing, great brands.

[00:05:12] And they did that originally by building a platform, like a design ops platform, essentially where work can get briefed.

[00:05:21] And this is what we run now, so that work can get briefed, designers can get assigned, reviews can happen, billing can happen, timelines, project management, very specific for kind of mid-market and enterprise segment.

[00:05:33] So not at the kind of fiver end of things where you're more of a marketplace, it's a tool and a platform for out kind of in housing a design team to so we can be an extension of a design team.

[00:05:48] And because of that, we've kind of been weirdly set up, but by happy accident, to be very interoperable with the Gen.AI world.

[00:05:59] And if you kind of abstract from what the Superside platform in itself does when a company works with us, is that there's a series of forms and text boxes that ask for your intent.

[00:06:11] And we guide your intent through, you know, something that we've worked on for a very long time, which is like, how do you take a good brief, creative brief or design brief from someone?

[00:06:21] Because that's hard in and of itself.

[00:06:23] That's a hard problem to solve.

[00:06:26] But I think we now have, I think about 120,000 projects that have gone through that platform over time, all centralized, all known from what the client has asked for to what they got back to what the revisions were to what the sign off is.

[00:06:45] And so that's kind of set us up with some pretty strong foundations to then look at, okay, well, if we were going to plug AI into any part of these processes within the intent process, for example, the briefing process, within the scoping process of like, how do you unpack that from client language into creative language, which is, you know, there's a translation layer needed there through to how do you then hand that off into something that you can do something with it?

[00:07:11] Like, you know, like a Figma or Photoshop.

[00:07:15] And so there's really just, you know, if you look at that pipeline, look at that workflow all the way through.

[00:07:21] If it's relatively centralized and you have relative control over where the data is and what you're doing with it and what UI you can build around it, then this service, which is essentially what it is, you know, people come to us, they want design services.

[00:07:37] Can be quite heavily augmented in quite a lean way with, you know, a large language model or an image model or, you know, different things coming in.

[00:07:46] And we're at different points of maturity in that at the moment.

[00:07:50] So some things are still very manual and some things are getting towards being quite automated.

[00:07:55] But yeah, in and of itself, Superside is a creative services company and we serve, you know, very large tech companies.

[00:08:03] That's kind of our particular niche.

[00:08:06] Great. I'm sure we'll go into much, much more detail.

[00:08:09] Let's talk about you a little bit, understand your own journey because you're a bit of a peculiar designer.

[00:08:15] You've been focused on creative technologies.

[00:08:17] You consider yourself a creative technologist.

[00:08:20] Why did you pick that and how did you end up in that kind of domain?

[00:08:23] Didn't pick it.

[00:08:24] This is the thing.

[00:08:25] It kind of found me through multiple iterations of a career.

[00:08:29] So I think I actually go back to my old university sometimes and give a workshop on how to find divergent careers out of what you study.

[00:08:41] Because I studied music at university.

[00:08:44] I'm kind of a relatively reformed hacker of kinds.

[00:08:51] Like spent a lot of my youth programming, a lot of my kind of higher education stage making and creating, but always veering towards the technical.

[00:09:02] And then during university, I met someone that worked in special physical special effects for like media, film and TV.

[00:09:13] And then I started just like wiring things up in his warehouse.

[00:09:18] I was like really the lowest of the lowest entry points to an industry.

[00:09:21] But it was an amazing industry, like the kind of wider entertainment industry.

[00:09:26] And so I just started making things and then making things in this creative way.

[00:09:31] So like kind of a bit of creative engineering, whether it was programming, whether it was actually physically making them.

[00:09:36] And that's really what I think a creative technologist is, is someone that can quite quickly pick up a skill set to a medium level of fidelity.

[00:09:46] And that they've got a reason, it's a jack of all trades in a lot of ways.

[00:09:49] They have a reasonable understanding across multiple domains of like physicality.

[00:09:53] So you can do some physical prototyping, you can do some software engineering, you can do some design.

[00:09:58] I generally say that's kind of my, my shtick is I'm an average designer.

[00:10:05] I'm an average developer.

[00:10:07] I'm an average creative.

[00:10:08] You wouldn't hire me for any one of those specific jobs.

[00:10:13] But what I am good, I'm, I am good at being is a voracious learner and, and a very like strong prototyper.

[00:10:21] So I can convey a new idea and something that hasn't happened yet through making it rather than going, here's a deck of what we could do.

[00:10:32] I generally will just jump straight into prototyping.

[00:10:36] And so I did that, you know, went from a career in special effects, which is the first kind of half of the career.

[00:10:44] Then into creative agencies, which is a very similar thing, but I just didn't even know they existed.

[00:10:49] I didn't know the job title creative technology technologist existed.

[00:10:53] And I didn't know the creative industries existed in the same ways within creative agencies.

[00:10:58] I didn't know that existed.

[00:10:59] So it was all just happy accidents of following what interested me.

[00:11:05] And also what was the most interesting emerging technology going on at that time.

[00:11:09] So it could have been augmented reality or real-time 3D and then moving into generative AI.

[00:11:14] But always with a very practical mindset of like, what can we do with this and how do we build something around it?

[00:11:20] But I wouldn't know necessarily how to say you should study this to be that.

[00:11:26] I don't know if that's necessarily the quality of a creative technologist anyway.

[00:11:30] This Design of AI episode is brought to you by PH1, a research and strategy consultancy that helps clients build AI products that customers want.

[00:11:39] Contact them about product discovery research to answer critical questions about what to build.

[00:11:44] Competitive analysis to find out how to gain an advantage.

[00:11:47] Service and customer analysis to identify the best use cases and value drivers.

[00:11:52] And workshops and concept testing to validate what will work and to fine-tune products.

[00:11:57] PH1 has worked on products for Spotify, Microsoft, the National Football League, Dell, Mozilla, Bell, and various health and higher education groups.

[00:12:06] Bring in an expert to make sure your products and teams are focused on what customers want.

[00:12:10] Visit their website to book an intro call.

[00:12:14] PH1.ca

[00:12:17] I think people forget how amazing the early knots were.

[00:12:21] They were a time when all these new industries were popping up and the people that could balance and jump across them really found their opportunities.

[00:12:27] If something happened with the advent of SaaS where we all thought there was a single pathway.

[00:12:33] Yes.

[00:12:34] And we lost sight of just how many amazing things were happening.

[00:12:36] But maybe this is that new genesis.

[00:12:39] I came from architecture and I remember seeing this talk from someone from agency and he was showing me what you can do with 3D.

[00:12:46] And I was just blown away.

[00:12:47] I had no idea this was possible.

[00:12:49] Yeah.

[00:12:49] And maybe this is what's happening now.

[00:12:51] But to that point, has there been some sort of advice or mentorship that really helped guide you to the point you're at now?

[00:12:59] I mean, the...

[00:13:01] So just actually quickly on that one thing that you just said.

[00:13:04] I do think that there is a new redefinition going on right now.

[00:13:08] And so people that can program can visualize now and people that can visualize can program now.

[00:13:15] So there is another inflection point I think coming.

[00:13:18] It's either happening or it's happened with this redefinition of categories of jobs is going through another kind of mesamorphosis right now.

[00:13:29] But to that point on how to thrive or how to find those opportunities and what advice I would give, you know, the constant curiosity.

[00:13:37] I know it sounds like a cliche, but I wake up in the morning excited to see what's happened, you know, been published the day before.

[00:13:47] And I find the advent of new technologies and new techniques incredibly interesting.

[00:13:56] And so they normally spark to me like a little web of possibilities.

[00:14:01] If that could do that, then it can do that.

[00:14:04] And I wonder if I plug that into that, what would that do?

[00:14:07] And I don't think you can fake curiosity, but you can definitely nurture it.

[00:14:14] So you can definitely try and go from, okay, instead of being just the consumer of material and just sit there.

[00:14:22] I get inspiration and information from social channels.

[00:14:25] But you need to be able to then turn that into behaviors that you act on it and you do something with it.

[00:14:30] And engineers in general and designers in general are pretty good at this themselves anyway.

[00:14:36] A lot of engineers will do a spec project or they'll do a personal passion project or they'll find a way to learn a new framework through the lens of making something.

[00:14:46] And the same with designers.

[00:14:47] They will do something similar.

[00:14:49] My advice to people that would want to bridge that gap is to then use something that's new to do something that you can't already do.

[00:14:59] So if you are an engineer, you need to work on more of the visual storytelling or you need to work on user experience design.

[00:15:08] Use a new curious novel technology as the vehicle to further your curiosity in that space and publish.

[00:15:17] I've got lots of opportunities in work by just showing and doing and talking about it and talking about it before it's ready and going into the community and having conversations about this.

[00:15:29] Your own personal bias going towards making rather than consuming is the kind of core behavior I think is that I generally see when I hire people that I want to see in my team.

[00:15:42] And it's always served me well.

[00:15:44] That's, I would say, probably my key piece of advice is make stuff.

[00:15:48] A couple months ago, we had a fellow Brit, Nick Sherrard, on who's an innovation strategist.

[00:15:53] And his analogy that right now is a time of, what was it, sea captains, people who have a vision.

[00:15:59] They know where they need to get to, but have no idea how to get there.

[00:16:01] And then pirates who are basically building this shit that we're all moving on.

[00:16:07] And a lot of times it doesn't always work out.

[00:16:09] You don't always hear about them, but they're the ones actually running out and being intrepid and crazy.

[00:16:14] Is that what you're saying too?

[00:16:15] Yes. I mean, that's literally how we describe our team internally in the business.

[00:16:20] We are a bit of a bunch of misfits, definitely a bunch of pirates.

[00:16:24] I think the one common theme is like a discomfort with the status quo.

[00:16:29] So, you know, it doesn't mean they're an easy team to manage.

[00:16:33] You know, if everyone is constantly questioning why, is it like, why do we do it this way?

[00:16:38] And is that the right way to be doing it?

[00:16:40] You have to have a certain level of resilience to be able to put up with those questions as a larger team over time.

[00:16:46] But pirates or like the speedboat ahead of the container ship, the people that are going off to each little while and going, actually, there is some resources here.

[00:16:54] We should steer the container ship this direction.

[00:16:58] You know, innovation teams in general or creative technologists that work within those functions.

[00:17:03] They're really good at that sort of thing.

[00:17:05] I will just say one thing on the downsides of that is you have to then also be really cognizant that you're not just flitting between kind of whatever the soup of the day is in terms of technology.

[00:17:17] And you're just going, actually, I'll be, you know, there is a bit of a cliche of like all the people in general say, I now used to be in crypto and they were other things beforehand.

[00:17:26] That is probably true on the kind of LinkedIn influencer side of things.

[00:17:32] You know, people that just want to have clicks and likes.

[00:17:36] But so you have to kind of balance against those two things.

[00:17:38] You want to be the speedboat ahead of the container ship.

[00:17:41] You want to be finding where to go and helping the sea captain go.

[00:17:46] Actually, there's stuff over here.

[00:17:48] But you also want to see some things through to like relative completion as well.

[00:17:52] And that's the thing that you have to kind of counterbalance within your own personality.

[00:17:57] Do you recommend this approach to people working in-house in a large org?

[00:18:00] Do you think they should be intrepid?

[00:18:02] Do you think it's better to be this wild and crazy misfit on the outside or in this room inside?

[00:18:08] It depends on the org.

[00:18:10] So, you know, a lot of larger companies, you know, they're really geared towards safety.

[00:18:16] I find an easy kind of, it's not mathematical, but it's quite easy to see within a business, like where the weight of voices come from.

[00:18:24] So if you have some departments, which are essentially just risk management departments of HR and talent and legal and, but not everything is legal by name, right?

[00:18:34] There's a lot of departments inside of companies which are essentially trying to de-risk situations.

[00:18:41] And if there is a larger balance of those departments within a larger organization, no matter how intrepid you're trying to be, you're probably going to be outvoiced when it comes to key decisions.

[00:18:53] Like the innovation team can just be like the infinite money pit sometimes in those large companies or the, you know, the department of shiny things.

[00:19:00] And you sometimes do something that is, yeah, all scissor window stake.

[00:19:05] I know a lot of creative technologists that are in those areas and they don't generally last because it sounds great to be innovation lead or creative technology lead at company X, not specifically X, but like, you know, as a company X.

[00:19:22] But then they normally go out now, the, the innovation by name, but not by practice.

[00:19:27] Let's help people out here.

[00:19:29] What are signals or flags?

[00:19:31] If I'm a product designer in house, but I want to push the boundaries on my team, what are some flags or indicator that this culture might appreciate it?

[00:19:42] The founders probably still in the business most likely is a really good indicator.

[00:19:47] So if you, if the founder is still within the business, then, and that founder is a, one of those profiles that gears towards risk and gears towards something that's absolutely right for the business.

[00:19:58] And you can find a way to not necessarily, you have to report to that founder, but you have to find a way to have a relationship within those kinds of core decision makers.

[00:20:07] If it's a professional, professional CEO, you know, career CEO that goes between, you know, really large companies, your likelihood of influence is less in my opinion.

[00:20:20] So founder led companies, I think is a very clear indicator.

[00:20:26] And if I'm in a mid cap SaaS company, how would I know that I'm not going to end up on the outs?

[00:20:33] I can't be certain in that area.

[00:20:36] And also when it comes to a SaaS company specifically, if it's going right, it shouldn't have an innovation team in a lot of ways because you should be innovating the entire product.

[00:20:49] It depends on how diversified the businesses in general.

[00:20:54] Yeah.

[00:20:54] I don't know if I actually have better advice for that.

[00:20:56] Like I only know what I know.

[00:20:58] Number one, what I won't work for in the creative space is a founder slash CEO slash creative director.

[00:21:09] If they, if that's all those three, three things in one, especially the creative director part with the founder part, they'll probably be a megalomaniac.

[00:21:17] They just want this kind of control over every aspect, but a founder that is really kind of risk aggressive, but it's willing to work.

[00:21:25] And like kind of actually give, you know, runway to their team.

[00:21:31] That's such a good signal of a company to go work for that actually has innovation.

[00:21:35] Like that is not going to have headwinds against innovation.

[00:21:38] And so that's what I look for is actually, is there the kind of fertile ground that we can do this here?

[00:21:45] Or are we in some sort of incremental company that's very risk managed, which is fine.

[00:21:50] I've rarely found that companies can be both things at the same time.

[00:21:53] That's fair.

[00:21:55] So you live and breathe new capabilities.

[00:21:58] That seems to be where you get your energy from.

[00:22:01] If I were to map out the last 10 years and the most impactful new capabilities that came out, how many of them are in the last two years?

[00:22:11] 93% of them.

[00:22:13] I guess it's so many of them, you know, it's so many of them genuinely.

[00:22:19] And I can speak to this quite directly to the point of, I've always been super interested in something new.

[00:22:28] Like, you know, augmented reality or kind of spatial technologies.

[00:22:32] I'm very interested in, very interested in number one, because I think it's the most natural method of interaction.

[00:22:40] Most likely over time is in, I'm really used to being in an environment that I can pick things up and, you know, information is coming from me spatially.

[00:22:50] And I've been trying to push that, you know, from my little area for a long period of time.

[00:22:55] But we've always been trying to find the use case in search of the technology in a lot of ways.

[00:23:02] Like the technology has existed, but everyone keeps on going, like, what do I do with it?

[00:23:07] And whether that's within my particular niche of, like, then within kind of creative marketing side of things.

[00:23:13] At the opposite end of the scale, in the last two years, you know, if you kind of maybe three years-ish,

[00:23:21] I can't even write down the amount of use cases I can think of for large language models and action models

[00:23:29] and how you can start to look at the generative technologies in general.

[00:23:34] I can't even, there's not enough hours in the day to think of the use cases.

[00:23:38] There's so many, and it's completely diametrically opposed, where I'm not even trying,

[00:23:42] I'm not trying to convince people that there is a use case.

[00:23:46] Now I'm trying to convince people that we can do it.

[00:23:49] Like there's a different, a different thing there.

[00:23:52] But it really is like a, for someone like myself, it's like someone's opened up the doors to the sweet shop and said,

[00:23:59] yeah, no, everything's here now.

[00:24:01] Go, go for it.

[00:24:02] Use the analogy of a door.

[00:24:03] So you're opening the door and Gen.ai is standing there.

[00:24:07] What is your relationship with Gen.ai?

[00:24:09] Why do you want to work with Gen.ai?

[00:24:12] Broadly at first, another area that fascinates me is brand.

[00:24:18] And brand is this kind of abstraction of a company.

[00:24:23] It's an abstraction of an entity.

[00:24:26] A brand in and of itself is something that it is, but it's something that it isn't.

[00:24:31] It's something that it works in context to the market around it.

[00:24:35] It's a quite ephemeral thing to capture what people feel about a brand.

[00:24:40] It's the promise.

[00:24:41] It's the trust that, you know, it will make you pay more for product A than product B because it's got a better brand.

[00:24:49] You know, it's just a very interesting thing.

[00:24:51] And so what kind of specifically fascinates me is we've been creating for brand by looking at, especially in like more production-y centered brand work.

[00:25:07] Is in this on this screen here, there's a text document that has a brief on it.

[00:25:14] And on this screen here is a PDF document that has my brand guidelines and rules on it.

[00:25:19] And on this screen here is a template of stuff that you've done before that's slightly out of date.

[00:25:24] And we've all agreed that this is the way that we get kind of creative work done that a marketer will send a brief and a brand agency would have compiled a brand rule book.

[00:25:37] And none of those things connect.

[00:25:40] And none of those things connect to the outside world.

[00:25:42] There's no cultural context coming in.

[00:25:45] There's no pulse on what's going on within like the here and now coming in.

[00:25:49] There's no additional connection between those three things, which is the core components of creation when it comes to brand work.

[00:25:55] It's the brief, the rules, and then the software.

[00:25:59] So what fascinates me about Genial is if you map a brand in a different way, rather than in layout and then in plain text and in a PDF, which is a desktop publishing format, era file.

[00:26:10] You know, if you map that in a different format, like a knowledge graph or, you know, something that could be looked at from different angles.

[00:26:19] And then you create different personas of a large language model or just a system prompt that can then just go and look at that information and come back with another piece of information.

[00:26:30] I think we're going to get to emergent properties around brand because that brand can be structured in a different way.

[00:26:39] And then looked at from multiple perspectives, like really, really elegantly and quickly.

[00:26:45] What I find fascinating about that is that I think it will be able to do stuff that we don't even know it can do right now.

[00:26:53] And that's the reason that companies go to agencies like big companies go to a creative agency.

[00:26:59] It's not because they want them to articulate their brand guidelines really strictly is that they want an additional point of view on the world added to their creative.

[00:27:07] They go to an external company because they need that cultural insight.

[00:27:12] And we've never done that with technology really well before, where you can go to this external entity that will help you with your ideas for your brand or it will create or abstract those ideas for your brand.

[00:27:26] And so that's what I find truly fascinating is that brand is just a fascinating concept by itself.

[00:27:33] So to put this in context, what you're trying to accomplish is what design ops has done, but to take that to another level, right?

[00:27:40] Yes and no.

[00:27:41] The excitement about it for me is we don't know how far this can go.

[00:27:46] And I think there's going to be some really exciting emerging properties coming from this AI plus brand space over time.

[00:27:53] I think there will be, it's a very rich like vein to mine.

[00:27:59] And I think we're only just getting started and that's what's exciting.

[00:28:02] Quite specifically, what design systems have done for product, I want to do for brand.

[00:28:10] Yes.

[00:28:12] But more so that I want that system to be able to actually make things itself rather than a design system is a store of rules.

[00:28:20] It doesn't generate anything new from it.

[00:28:23] Yeah.

[00:28:23] Design system is something, it's mainly a communication system.

[00:28:25] It's like a, you know, it's handoff between design and developers or it's documentation.

[00:28:30] So people know we use this, but we don't use that.

[00:28:34] We use this, we don't use that.

[00:28:35] Which is helpful, but it's constraints in a lot of ways.

[00:28:39] And constraints are good.

[00:28:41] I want a design system or brand system that can make its own work off the back of it.

[00:28:47] And also that within the context of the culture that it's within, not just within the context of the rules it sets.

[00:28:54] I think there's some really interesting territory there.

[00:28:57] So what would you consider the job to be done for Gen AI from the perspective of creative teams?

[00:29:03] What is its allyship in that situation?

[00:29:06] So I've been asked that question like multiple times and we've tried to unpack it.

[00:29:10] I actually don't think the job to be done is any different than the human job to be done, really.

[00:29:15] So if you look at it at quite a broad level, it's, you know, why does marketing, comms, you know, brand communication, marketing, advertising exist?

[00:29:24] You know, it's generally to shift the bottom line and to sell more things, right?

[00:29:30] That's the thing that, you know, if you go all the way up the top, but the job to be done of I need to be able to make some display ads or I need to be able to make a campaign that goes across these channels.

[00:29:45] You know, you can, you can articulate that in different ways, but I don't think it changes.

[00:29:49] You're still trying to do the same core thing, which is you're trying to get your brand cut through.

[00:29:54] You're trying to increase click through.

[00:29:56] You're trying to do all of the kind of core metrics that you're trying to make is that if you can increase the speed of execution across multiple dimensions of that.

[00:30:09] As in like, so if you can increase the speed of production from two weeks to two hours, for example, what more can you then do?

[00:30:19] Number one.

[00:30:20] So do you use that rest of that time to increase quality or change format?

[00:30:28] Format expansion is a very interesting one.

[00:30:30] Just in general, can you do more things within that space or can you be much more tapped into what's going on within the market?

[00:30:39] So you can be much more responsive to what's got, you know, so, you know, generally there's a constraint on creative output.

[00:30:47] Marketers want to be able to publish much more often than they can produce generally.

[00:30:53] So there is a bit of a right sizing there of that demand, like what marketers need to what creatives can produce.

[00:31:00] So I don't see the job to be done as any different.

[00:31:02] I see it as accelerated and I see it as augmented.

[00:31:05] So you can do more things and things to a, to a higher quality.

[00:31:13] And I'll give you one example on that.

[00:31:15] The super side marketing team a couple of months ago made a Gen AI music album about creative problems.

[00:31:24] So there's like a song called, can you share the Figma link?

[00:31:27] And there's just kind of very niche designer focused problems.

[00:31:34] And it was written with a large language model from the lyrics, obviously altered and thought about by the creatives.

[00:31:42] It was then we put it into Suno to like then create the music and then did an edit from that.

[00:31:48] We then went out and shot a couple of music videos and made a couple of music videos in runway as well.

[00:31:54] We then published it on Spotify.

[00:31:56] There's a full Spotify album of it.

[00:31:58] And then we made a microsite about that.

[00:31:59] And I think we, the end to end process between someone having an idea saying we should probably do this to it being live was two and a half weeks, roughly.

[00:32:08] 10 songs on an album, music videos, cover art, all of the things.

[00:32:13] And the reason I give you that analogy is could we have done that without AI?

[00:32:19] Yeah.

[00:32:20] We could have potentially all written songs and written the lyrics for those songs.

[00:32:24] Or we could have commissioned some songwriters to do that.

[00:32:27] We could have gone to a music studio.

[00:32:28] We could have got a producer.

[00:32:31] We could have then gone out and done the music videos.

[00:32:35] And that would have taken us four months, probably.

[00:32:39] Which means we wouldn't have done it.

[00:32:40] There's no way our CEO would have signed it off.

[00:32:43] There's no way that we would have, like, we wouldn't have spent the money.

[00:32:46] We're a scale up, right?

[00:32:47] So there's no way that we would have done that.

[00:32:50] And now we will do different things because of it.

[00:32:53] So that format expansion is the same job to be done.

[00:32:56] The job to be done is that we want to raise awareness.

[00:32:59] We want to be able to create materials.

[00:33:00] We want to be able to publish them.

[00:33:02] But now we just have different ways of going about it.

[00:33:05] And I think that expanded possibility, I loved seeing that for my marketing team because it was entirely from themselves.

[00:33:12] It's beautiful.

[00:33:13] But it also makes you wonder, everything we're talking about today is about outputs and efficiency gains.

[00:33:19] But it really feels like your mandate is culture change.

[00:33:23] It really feels like what we're actually here to talk about is what designers should be doing in this liminal moment where, in reality, we should be embracing change in some form.

[00:33:33] So what would you say to designers as they face this impending culture change that perhaps they're not aware of?

[00:33:40] So we started our AI journey with just two of us, me and my colleague, Alexander.

[00:33:52] And we didn't know when we started where things would go.

[00:33:57] We started by just shadowing projects and figuring stuff out.

[00:34:02] And I would say for both of us, he's another creative technologist.

[00:34:07] We're both average people.

[00:34:09] We don't have this sunk cost fallacy of I'm an expert at anything.

[00:34:13] So if I've spent 20 years of After Effects crashing on me or Cinema 4D or whatever your flavor of software and tooling and hardware is,

[00:34:24] you've gone through multiple kind of pieces of pain as a creative from higher education to your internship to your laptop running so hot and crashing and then your file not saving.

[00:34:39] You know, the way that you work is really intrinsically kind of quite deep within your identity.

[00:34:46] And the real experts find that very difficult to let go for good reason.

[00:34:50] And I don't think that's in any way of a challenge.

[00:34:53] All I'm saying is I don't have that challenge because I'm just not good enough or anything.

[00:34:56] So I'm quite willing to let stuff go because I'm like, oh, cool.

[00:34:59] It can do that now.

[00:35:01] That means I can focus on all the other things I want to do.

[00:35:05] So I'm just going to just want to say that up front is that I can't speak from a pure expert's point of view because a pure expert may have a different opinion to me.

[00:35:14] So it started with two of us.

[00:35:15] We expanded that out when we found some fertile ground to 20 of us.

[00:35:19] We've trained roughly 240 creatives and designers now.

[00:35:22] So we've got a reasonable, significant amount of people that have gone through this journey with us to see where the broad trends are of like the people that are really, really good and really, really adept.

[00:35:31] The ones that find it a bit more difficult and the ones that will find it even more challenging over time.

[00:35:36] And the ones that are really good.

[00:35:39] Number one, they're kind of.

[00:35:42] Are very willing.

[00:35:43] They do have a high tolerance for pain, right?

[00:35:45] They have a high tolerance for bad software, like shitty UX things being in discord.

[00:35:53] And they're just quite willing to look through the pain of the software to be able to get to the end result that they want to get to.

[00:36:02] And that's the other key point is that they have a really clear point of view.

[00:36:06] They have taste.

[00:36:07] They have discernment.

[00:36:09] They have good visual acuity and they can articulate what they want in words or in different mediums.

[00:36:16] And they also can have really good judgment on what comes out the other side, if it's good enough or not, if they don't like it or not.

[00:36:22] And I see this all the time on LinkedIn.

[00:36:26] I'll see people going, isn't it amazing what software X, Y, Z can do?

[00:36:30] And they just show something that is just like eyeball bleeding.

[00:36:33] It's amazing, but they just don't have any idea what amazing actually looks like.

[00:36:37] And creatives and designers in general have a great idea of what good enough is.

[00:36:43] And they have a really good way of conveying good enough.

[00:36:46] And so those people that are quite tolerant to forget the old way and go dive into a half-baked piece of software that is full of bugs and it's going to, you know, that takes a certain resilience.

[00:37:04] And if you pair that up with someone that is highly creative, like an art director or a copywriter, someone that knows the nuance of language or can really tell like why something works.

[00:37:19] Like, why do I like that image or why do I like this design?

[00:37:23] And they have a point of view.

[00:37:25] That's a really powerful combination.

[00:37:28] If you are essentially, and I mean this in the nicest possible way, like someone that works in design, but they're not really a designer.

[00:37:38] Or they're someone that pushes pixels, but they don't think about strategy or they don't think about the concept behind what they're doing.

[00:37:45] There is no higher level thinking.

[00:37:48] They're the ones that stand not less of a chance, but they're the ones that vibe with it less.

[00:37:54] Because, you know, they're very connected to the actual kind of real physical framework of I have a taskbar here and I have, you know, a taskbar on the left.

[00:38:03] And this is what I work on.

[00:38:04] I work on a canvas and I do this.

[00:38:06] Where for me, like I want to, if I make something, I want to move people.

[00:38:10] I want it to do something for them.

[00:38:13] I don't care how it's made.

[00:38:15] And so the people that have taste, judgment, discernment, a real point of view, they're the ones I see the most amazing work come from.

[00:38:24] And I've got specific examples of people I'm like, that's the kind of person that I want a hundred of those because they're just so wonderfully adept at it.

[00:38:32] Now imagine I'm a product leader or design leader in one of these big orgs and I want to change the culture.

[00:38:38] How can I do that?

[00:38:39] What are some steps that you might suggest to actually shift these designers and rattle their cages and get the product team to actually look at things differently?

[00:38:48] Well, so rattling cages is one approach and that's more of my approach.

[00:38:55] But I've also got some wonderful counterparts in the business that take a more people led approach.

[00:39:00] And I think between the two, I guess, lines of thought, we found something quite positive.

[00:39:08] You do need to create the psychological safety that you're allowed to do this.

[00:39:13] And that psychological safety has multiple components to it.

[00:39:18] Number one, it depends on your risk profile of a business.

[00:39:21] But if you are just going to lock down everything and say, no, no, no, we just use Copilot and you cannot sign up for the latest Flux model and you cannot go over here and sign up for this beta and you can't put a credit card there and you can't put our information there.

[00:39:35] There's a real kind of strange thing with image models and people being really scared of them with promptings.

[00:39:41] You don't prompt your brand, you prompt an aesthetic, you don't feed your brand's information into a prompt that will work.

[00:39:50] But people are terrified that they're going to get their secret sauce in these image models.

[00:39:54] That's just not the way it works.

[00:39:55] That lock down of tools comes from the top.

[00:39:59] And so what you can do is be like, okay, here's the three things that we recommend that you use.

[00:40:06] And then in a certain amount of your time or a certain amount of spend, go and find out of the 3000 tools that have been released in the last two months, go find the one that works for you.

[00:40:18] Because it's an amazing time to be a creative right now is that there's so many people making creative tooling and creative models and creative things.

[00:40:24] So don't dictate what they should use.

[00:40:28] Crowdsource it by creating that kind of buffer to say, no, you can do that.

[00:40:33] You can go and find those things.

[00:40:35] The other part of psychological safety is the kind of elephant in the room about, am I training my replacement, for example?

[00:40:43] And so to speak very directly to that, we have seen no evidence within our business that this leads to a lower value for the people that are highly adept at it.

[00:40:54] So we want more of those kinds of people.

[00:40:56] They are able to service our clients better.

[00:40:59] We're able to output more work.

[00:41:00] They help us lead innovation like that.

[00:41:02] We have hired because of AI, not anything else.

[00:41:06] And we're a company that values growth, right?

[00:41:09] So we want to, maybe if you're in a private company that's been taken over by private equity and there's a different vibe to it, then, you know, that's fine.

[00:41:17] But that psychological safety is paramount.

[00:41:19] Like you have to address and understand and hear people.

[00:41:23] And designers do have a very allergic and visceral reaction.

[00:41:26] You know, there is people that are really, really against these things.

[00:41:30] The other side of it is like some principles around what you stand for when it comes to AI within your company.

[00:41:40] And we're only just getting to the point now.

[00:41:43] We didn't start with the principles.

[00:41:44] We found the principles along the way.

[00:41:46] You know, you can either set them out as complete, you know, this is what we do.

[00:41:50] Or you can say we are going to find our principles along the way.

[00:41:54] But from our point of view, like we want to meet creatives where they are.

[00:41:59] So we want to make sure what we build is we're building for the creative community that we've come from.

[00:42:03] So we want to make things, if it's plugins, if it's data pipelines, if it's models, if it's automation between software, that is going back to things like Figma or things, you know, where designers love to be.

[00:42:17] There's an amazing community between Figma, the product and Figma, the design community that people actually genuinely want to be there.

[00:42:25] That's gold dust.

[00:42:26] No one gets that right.

[00:42:28] Like, so we should embrace that.

[00:42:30] And then the other side of it is, you know, if you are a brand and company, your first party data, the things that you've made, the things that are not present in the model, the things that are not present online, maybe because they're within your own internal infrastructure.

[00:42:46] That data is your voice.

[00:42:48] That data is what makes your brand or your creative genuinely yours.

[00:42:53] And you should always value first party data, like what you make humanly.

[00:42:58] And then you amplify that with AI.

[00:43:00] Don't try and replace the first party data because that's when you get to a point of, you know, it's all homogenous brands and it's all homogenous products.

[00:43:09] So first party data and creators where they are, like, you know, is some principles that we're working towards, like really making that concrete.

[00:43:18] And the psychological safety of, like, it's okay to try, it's okay to fail.

[00:43:21] We've got loads of kind of systems really within the business of, like, okay, if we screw this up and it takes us twice as long as we've billed for, we will absorb that cost.

[00:43:31] It's okay because we know we're on the edge right now.

[00:43:35] We don't make that someone else's problem because we've asked them to push.

[00:43:38] So those areas are really important to get right at the beginning and set the tone because then you start to see these amazing things that come from that.

[00:43:46] You start to see people finding, I've got, like, these, like, you know, really young people that work that I hired that are just then starting to kind of do stuff that I can't do.

[00:43:57] I'm like the old man in the team that I can't, you know, I remember back in my day, I could just use Mid Journey.

[00:44:02] And now they're doing all these different things and, like, they're doing all these, you know, and it's amazing to see.

[00:44:06] So it's lovies to see those kind of flowers bloom.

[00:44:10] So we've talked a lot about innovation and culture change.

[00:44:13] I'd like to get into specifics of what you see the future looking like.

[00:44:17] And something that I personally get caught up a lot on is this idea of bringing ideas to market quicker.

[00:44:25] It necessitates knowing what good is.

[00:44:27] It necessitates interpreting that documentation, those constraints and such.

[00:44:32] How do you define and quantify what good is?

[00:44:37] Fortunately, we're not a pure product company.

[00:44:40] I think we just before we started, we were kind of talking about there's some constraints that like, say, for example, Adobe come out with a new thing.

[00:44:47] You know, Figma come out with a new thing.

[00:44:48] Figma come out with make design.

[00:44:51] It kind of creates a load of quite negative ripples in the design community.

[00:44:55] And they pull it and they're coming back with a thing called first draft and that's looking really cool what they're doing.

[00:45:00] But they have to bake those constraints into the technology.

[00:45:04] You know, they have to kind of slightly neuter what's possible because if it goes too far in any direction, it's quite a lot of reputational risk.

[00:45:11] It can do some quite a lot of damage as a product company that can, that gives quite a lot of generative ability to its users.

[00:45:19] And what I'm saying that is because we don't generally have that problem.

[00:45:21] I don't actually have to think about the question that you've just asked because we've got humans that do that.

[00:45:26] My job is to help that human with better decisions and accelerations of different parts of the workflow so they can do that particular part that they are really adept at.

[00:45:40] So humans are really good at going, it's not that one, it's that one.

[00:45:43] And it's that one because, and we should double click into that one and do more of that one.

[00:45:48] I actually don't know.

[00:45:50] And I haven't thought very deeply about how you replace that.

[00:45:52] And I know you can technologically look at that in terms of, okay, we could have some sort of vision model that is going to unpack that into this.

[00:46:00] And we can look at this, the pipeline of really auditing what's on the page.

[00:46:05] But I still think it would do a worse job than a human would because a human has way more emotional context to why something is resonant.

[00:46:13] How do we visually brief a human?

[00:46:15] So for example, like the main part that I'm working on is how do you go from brief to first creative draft?

[00:46:21] And first creative review in order of magnitude time.

[00:46:24] So a company wants, you know, some UX design or they want digital ads making the brief super side.

[00:46:34] And I want to be able to turn that around and measure it in seconds rather than measure it in days.

[00:46:39] Like that's the ultimate aim.

[00:46:40] But what it has to come to is if I know everything about that brand, if I have their design system.

[00:46:47] If I have all of their past context of what they've done with us and I have their brief, can we transform that information and basically write the code of the design onto the page within Figma?

[00:46:59] Write out the requirements.

[00:47:02] Say like, actually, they want this is the headline and they want this here and they want to see these kinds of images.

[00:47:06] And we just write that out.

[00:47:07] We give 10 options on the page.

[00:47:11] AI won't come in and go, you should use option one.

[00:47:14] It may, you may be able to do some heat maps going option one may work better because it has a bigger CTA or the, you know, that it has a person on the image and you know, that's a, that's a good thing for people's attraction to it.

[00:47:25] But a human will come in and go actually definitely not option four, five, six.

[00:47:30] Let's take them out, delete those.

[00:47:31] Let's combine those two things.

[00:47:34] Let's take them out.

[00:47:34] And then they'll like go in and fine tune it.

[00:47:36] They'll tweak it.

[00:47:37] That means that they haven't had to go off and find the file or like think about it, or they've got other, other distractions going on.

[00:47:44] You're empowering that human decision-making rather than trying to replace that human decision-making.

[00:47:49] I actually haven't really thought about how to do it the other way around.

[00:47:52] Like, I just don't know how to do it the other way around because it's quite difficult.

[00:47:56] So, so what types of workflows does this work quite surprisingly good at and what types of workflows is it surprisingly incapable of matching a very manual process?

[00:48:10] Types of workflows is very good at.

[00:48:11] Firstly is where the value is.

[00:48:13] If something is going to live in the world for two weeks, truncating that process to make it is highly worthwhile.

[00:48:21] So if it takes you four weeks to make something that lives in the world for two weeks, like an email marketing campaign, for example, you know, email marketing lives for a day.

[00:48:31] You know, generally the relationship between how long it takes you to make it to how long it lives for, I think is super relevant because you, you know, the effort you should be putting in should be proportional to the impact it will make.

[00:48:44] So things like email marketing campaigns, performance ads, digital ads, social media posts.

[00:48:49] Anything that is like kind of downstream underneath the brand within the marketing space, I think is incredibly relevant.

[00:48:56] Anything that is a mixture between layout, copy and image, all of those things can be entirely augmented.

[00:49:04] Layout is in like where your logo is, where your brand headline is, where, you know, what your padding is, the template something sits within.

[00:49:13] That's just code at the end of the day.

[00:49:15] You can unpack that very quickly into, you know, react components.

[00:49:18] You can unpack that very quickly into a design system in any way.

[00:49:23] Short form copy of like, you know, your headlines, your body, your, you know, your sub headlines entirely possible.

[00:49:30] And your images, you know, rasterized graphics, photography, 3D style, illustrative style, completely possible.

[00:49:38] And if you combine those elements together, you can pretty much get to a very broad swathe of the creative output a company needs.

[00:49:46] So they'll need blog headers.

[00:49:48] They'll need blogs.

[00:49:49] They'll need emails.

[00:49:50] They'll need presentations, which are presentations are layout, copy and images.

[00:49:55] You know, they, they are all of those components.

[00:49:58] So if you look at it, just in those components, you can pretty much get to most outputs.

[00:50:03] You can also do that with like web, right?

[00:50:06] Or app design or product design.

[00:50:09] The only thing I would argue slightly against it, it kind of turns into a bit of a party trick sometimes when you go, okay, I'll do text to UI.

[00:50:17] And lots of people are focusing on text to UI products at the moment where you can see this little magic trick of a page populating out and it will be writing the content and it will be doing the layout.

[00:50:28] Which is great.

[00:50:29] But if it's a web app that actually has to have anything to do, it actually has to do something rather than just communicate.

[00:50:36] It actually has to then connect to this thing and it has to be able to, you know, have any information coming in or out of it.

[00:50:42] You're most likely then going to spend the next three months developing that.

[00:50:46] So doing the party trick at the beginning, going from text to UI is great.

[00:50:53] If you're then going to spend, yeah, three to six months during the initial development, it doesn't really matter that you've made the UI in three minutes.

[00:51:02] It's a good, fun thing.

[00:51:05] But if you've got a thing that has to last for three hours in an email marketing campaign and you've made it in three minutes, there's a really good relationship there between what the material is intended for.

[00:51:17] And how long it takes you to make it.

[00:51:19] And the same with brand as well.

[00:51:21] If your brand is going to live, your logo, your lockup is going to live on the side of your building for the next 20 years.

[00:51:27] Or it's going to go in your Super Bowl ad.

[00:51:29] It's going to be seen by 200 million people.

[00:51:31] Is the additional efficiency worth it?

[00:51:34] And I just would argue it's not necessarily like what, why?

[00:51:39] Because you should be thinking about the longevity of your brand.

[00:51:44] Or if you're putting all of your heart and soul into what it looks like on your t-shirts and buildings and everything.

[00:51:51] I just don't know if there's enough effort and impact kind of correlation there to why you'd want an efficient process.

[00:51:58] There's other areas which I think we can all agree we want efficient processes.

[00:52:02] Things that we hope could make themselves.

[00:52:05] Is in like, we need some banner ads that need to go out tomorrow.

[00:52:09] And they're going to go and live there for like a week.

[00:52:11] Fine.

[00:52:12] Let's just do that with AI.

[00:52:14] I think there's a real difference between where these things go and what they're for.

[00:52:19] If we look in the next two years, the capabilities that you can expect to come online, how will that change your platform?

[00:52:28] Everyone's going on about agents.

[00:52:31] And agents doesn't really mean something, but it kind of means something in some ways.

[00:52:36] And if you think of an agent as a human, or I don't know, maybe want to personify a bit, as a kind of a decision maker within a chain.

[00:52:46] At the moment, the general challenge is that the accuracy and therefore efficacy of those agents to be able to complete a task to a sufficient fidelity.

[00:52:58] If it gets to 80% good enough, for example, in a certain task, and then it passes that task on to the next agent in the chain, which is 80% good enough.

[00:53:09] You're suddenly on a logarithmic curve of bad quality coming out at the end.

[00:53:15] And suddenly you're at like 0.2% very, very quickly of it being good.

[00:53:19] But then there's also the inverse is true.

[00:53:22] So if you're starting to get to like 98% accurate at the top end of what you're trying to get that agent to do, like the UI layout, for example, and then the copywriting.

[00:53:35] Going back to the UI layout, if you're writing code, it has to be accurate.

[00:53:39] Like you don't want to be then debugging that code every single time.

[00:53:42] So if you're writing code at a lot of efficacy, especially for the kind of mid-level stuff that we're doing.

[00:53:48] Then if you pass that down the line to the next agent, that will write out the copy that will go within, you know, the content that will go within that layout.

[00:53:55] If each one has a high degree of accuracy and that accuracy is enabled, in my opinion, by good retrievability from a retrieval augmented generation, most likely is stuff that's within the large language model and stuff that's external to the large language model.

[00:54:12] And how do you get the external data into the system context of the prompt?

[00:54:19] If those things have a good degree of reliability, then you can start to look at much more sophisticated long-scale processes.

[00:54:30] At the moment, we just do not have the level of confidence to say, actually, you want to go, we'll start this chain here.

[00:54:37] And it'll go all the way through down to a full final draft of a thing without going directly to the customer, maybe like directly to the client.

[00:54:47] Do you like this?

[00:54:47] Or do you like that?

[00:54:48] Or do you like this?

[00:54:49] Or do you like that?

[00:54:49] Put that in front of them if they brief.

[00:54:51] The rate of improvement in large language models, the amount of investment in that space, just in general.

[00:54:58] How easy it is now to fine tune image models.

[00:55:01] So Flux has made it incredibly easy to fine tune products and specific looks and aesthetics.

[00:55:09] One way machine, the video AI company, we've been working them for a very long time now.

[00:55:15] It was this infinite slot machine.

[00:55:17] We maybe get one in 50 would be right.

[00:55:20] And at the moment, I think it's like one in 10.

[00:55:22] And if we get that down to one in four, we're cooking.

[00:55:26] Like, you know, it was this really bad like roulette table before.

[00:55:30] And they just have a rate of improvement.

[00:55:32] So there's a rate of improvement in the LLMs, in diffusion models and in video models.

[00:55:36] And that enables then us to chain those together through multiple different agents that will say, actually, we'll go to this model to do that thing.

[00:55:43] Which means our ability to output quality, client ready, useful work improves a lot over the next two years.

[00:55:54] And at the moment, these are separate tools, separate people, separate processes, because we don't have a high degree of confidence that the middle part of the agent is not going to screw it up.

[00:56:05] And so I think the opportunities there are each rate of improvement then leads to more multi-agent processes.

[00:56:15] And multi-agent processes, I think, are going to be the most impactful for a business over time.

[00:56:20] To close this out, I want to think about that future product org.

[00:56:25] You know, whether this is two or five years from now, an organization that's fully bought into this.

[00:56:30] What are those roles on the team?

[00:56:33] More technologically focused?

[00:56:36] Is there still a designer?

[00:56:37] What does it look like?

[00:56:38] I don't know.

[00:56:40] And the reason I don't know is I keep looking at my team and how to best organize them.

[00:56:44] I have a team of 10 who are kind of researchers, technologists, designers and creatives.

[00:56:49] Do you put the technical people together?

[00:56:52] Do you buddy them up in little pods?

[00:56:55] You know, we've got a much larger company of 800 and we continuously have this conversation.

[00:57:00] I think there's more of a redefinition of titles and functions over the next five years than I think has been over the previous five years.

[00:57:11] So a product manager who may have been encumbered slightly on visual communication and slightly on technical.

[00:57:21] For example, depending, you know, most product managers will veer towards one of those tracks.

[00:57:25] They may come from design or they may have come from engineering or they may have come on a different path.

[00:57:34] I think a product manager will have more superpowers because they'll be able to communicate in higher fidelity.

[00:57:39] They'll be able to communicate because both up and down the chain they need to communicate to.

[00:57:45] So they'll be able to, you know, their words should propagate into design more as in like, can you make me mock-ups based off these things?

[00:57:52] And it should propagate that into development more.

[00:57:55] So I think if you think of those areas of the fidelity of communication within a product tool should become more sophisticated at the time.

[00:58:03] And then things which are some areas that may be more at risk, which are people within design systems, for example.

[00:58:13] If you spend a lot of your time writing to documentation rather than thinking about what the system does and why it's there.

[00:58:20] Like your job, you should be really trying to focus yourself on making sure what you're doing provides impact rather than it just keeps you busy.

[00:58:29] Because, you know, maintenance of a large design system within a product or for example, that's a big job.

[00:58:36] It may not be as necessary in the future when I think design system information will be stored and used in different ways.

[00:58:44] I think there's a really interesting area between design systems and large language models.

[00:58:49] I think they're incredibly interoperable.

[00:58:51] And I think they should be interoperable.

[00:58:54] But the short answer to the question, which I just gave the long answer to is, I don't know, because I think there's just that we are at this very fuzzy future stage now where there is quite a lot of role definition going on right now.

[00:59:07] The only specific answer I can give is people that are curious.

[00:59:13] And I have a bias towards this.

[00:59:15] People that are curious, people that play, people that test, people that prototype have a stronger likelihood of keeping their skills sharp and being more relevant.

[00:59:24] Hopefully you've got those kinds of people within your organization.

[00:59:28] Thanks, Phil.

[00:59:29] I guess last thing here.

[00:59:30] Are there any resources that you recommend people follow up on or places where they can see your experiments?

[00:59:37] Something that will help inspire them or guide them to become these intrepid people who have stronger futures?

[00:59:46] So, I mean, I guess there's all of the kind of normal that I will voraciously listen to of Lenny's podcast and I'll listen to yours and I'll listen to The Verge and I'll listen to all these audio sources.

[00:59:58] My newest way of capturing information is definitely the Notebook LM way, which is, you know, every week I'll build a new notebook and I'll use it to ask questions of the 10, 15, 20, 30 different things that I've found.

[01:00:10] And I've found that as a really useful way of, there's a lot of information overload at the moment.

[01:00:18] So, in fact, recommending new sources may not be a good thing.

[01:00:22] Recommending ways to cut through those sources may be the right way.

[01:00:26] And Notebook LM, I think it's the best thing that Google have come out with in a very long time.

[01:00:31] Not the podcast audio overview is fine.

[01:00:35] And it's a good party trick and I think I quite like it.

[01:00:38] But the using its chat function with citations back to the source material, I find super useful to actually keep me grounded in things that are very high frequency.

[01:00:49] And I'm not a Twitter user or an X user.

[01:00:51] I use LinkedIn and I actually find it to be the best community for this at the moment.

[01:00:56] I find it to be one of the more positive communities.

[01:00:58] If you can just surround yourself with a few interested builder folks on there.

[01:01:04] And if you search the term creative technologist and you click on those people's job titles within LinkedIn and start following them, nearly always they're posting interesting stuff on experiments they're making.

[01:01:16] And they're the people I seek out and follow, the makers.

[01:01:19] You've given us so many useful hacks.

[01:01:21] Thank you so much for your time.

[01:01:22] I'm really looking forward to see what you come up with because honestly, when I saw your presentation, I was like, this is going to change everything.

[01:01:30] Thank you for listening to the Design of AI podcast.

[01:01:32] We interview AI leaders and discuss the latest innovations.

[01:01:35] We help teams learn how to leverage AI to reshape their industries.

[01:01:39] If you like learning about AI and how to grow your career, make sure to follow us on your favorite podcasting app.

[01:01:44] It really helps us if you leave reviews.

[01:01:46] To get more AI and career resources, subscribe on Substack at designofai.substack.com.

[01:01:53] This episode was hosted by Arpe Drag Figueroero, the founder and head of product strategy for PH1.