Inside Google’s Creative Frontier with Josh Woodward & Robert Wong

0:00The really big unlock we found is you get a small group of people together, you let them kind of dream big and you kind of just let them kind of create and jam. Add the magic of human creativity to add a little bit of beauty, joy, meaning, humanity and soul to the products that we ship. Good products always make you feel something.

0:23Welcome to Mountain View, California. We're recording in front of 500 marketers on stage at Google Marketing Live, with thousands more watching via live stream. This episode of Frontier CMO is your front row seat to one of the most important conversations that matter today. How AI can make us bolder and more creative marketers. I'm joined on stage by my colleagues, Josh Woodward and Robert Wong, Two of the sharpest minds at Google, working at the intersection of engineering and imagination. Josh is the engine behind Gemini at Google Labs. Robert is the creative visionary leading Google Creative Lab.

1:00Together, they're pushing the frontiers of AI, using it as a true creative partner. If you're only using AI for speed and scale, you're missing the magic. Here's my conversation recorded at GML with Josh and Robert.

1:17Hello everyone. Welcome to Frontier CMO. This is a super exciting moment. Woo, Josh! Here with 500 of our friends. Robert, Josh, how are we doing? Great. Excellent. Okay. So I was working out backstage. Between the three of us, we have nearly 50 years of work experience at Google. That's a long time. I think that makes us like old hands. And for me anyway... Speak for yourself. I've never been more excited about the possibilities. All the technology, all the AI, all the stuff we've seen at GML, at I.O. It's kind of incredibly inspiring. So before we get going, and I'll do intros in a second, what's one piece of AI that has inspired you in the past week, month, six months?

2:04Go for it. Can I see three? In the last month, this is all just in the last month. I mean, every day I see amazing things, but the last month I was at Coachella. I saw that Enema set, the DJ set, where the background was all generated with AI. It was mind-blowing. And then I was in Vegas, and I saw The Wizard of Oz, like an old classic movie, but in this sphere, and all outpainted. It's just phenomenal what it could do. But one of the most inspiring things is I know a music video director, pretty well-known. I'm not going to say his name because the project is still in the wraps. And he's always wanted to make a movie, but Hollywood wouldn't give him a time of day, even though he's a very successful music video director.

2:43And so with AI and introduction of flow and stuff, last year he's been cobbling together and he just screened for me last week a 90 minute feature anime film. And it was mind blowing. So like one person in his apartment doing it completely by himself and without having, you know, Stuart has a green lighting it or anything. It's just mind blowing. So that's the stuff that's really exciting. That's cool. I guess one for me I'm thinking of, I don't know if any of you had a chance to play with Project Genie, which is some of this amazing tech that lets you kind of build a world and explore it. I remember exactly where I was standing when I saw that for the first time.

3:26But in the last week, we've now added the ability. It's hooked up with Google Street View. And you can put in an address, obviously go to that street, and then enter into it. And so with my wife and our kids, we've gone now to our hometowns, places we wanted to visit. And it's just unbelievable to be teleported like that. Okay. If I go into the prior versions of Google Maps, can I go to my childhood home and recreate that? Yeah, you should try it. Right now, it's hooked up with whatever is live in Street View. You literally can just type the address in Genie. It drops you there. You hit Explore, and you're there.

4:04Your days seem interesting. Okay. We built that together. That's right. This is a joint thing. We're going to talk about it. All right. So, Josh, inside of Google, you are a little bit of a rock star. Just tell everyone, what do you actually do? What are you doing all day long? What does it mean to be building the Gemini app? Oh, wow. Well, no day is the same. I would say it's some mix of trying to spend a percentage of my time really trying to think about and build things that we think are in the future. And we're always interested in things that are almost possible. This is when we do some of our best work together too.

4:38So when you get a new model like yesterday's 3.5 flash, that just reset the baseline of now the almost possible has shifted because a lot of things just became possible. And so part of it's that. Part of it, I spend a lot of time still with customers, developers, just people using our stuff. I'm always trying to find out what do they like, what do they not like, what might cause them to switch, what might cause them to churn. So I actually try to carve out a lot of time for that. And then I also read tons and use a lot of products. I'm trying to use everything and try to develop opinions about it.

5:12And then, of course, across Gemini, Labs, AI Studio, you can think of it as like a constellation of teams. The smaller, the better. A lot of our Labs projects are five, six, seven people. You kind of align on a mission, and you just let them cook. And we try to build to learn as fast as we can. So in a world of AI, you just said you read a lot, which might seem kind of surprising. So just what are you reading? Oh, well, I guess I have a mix. So I still will read books, hardback, paperback, Kindle, Audible, Libby, you name it. I have kind of a diet of that. I'm a lot on X and other places online, just trying to see where the trends are and like kind of where the frontier is.

6:00And then I also listen to a lot of stuff and try to ask lots of people. One of the great things, well, with people like you all, but also I would say at Google too, is there's just experts everywhere. And so I'm always trying to understand what kind of surprises them. And again, I'm always trying to kind of figure out like what's almost possible in a lot of those conversations. What's almost possible sounds like a pretty good thing for a T-shirt. So I like it. Robert, you run the Google Creative Lab. What does that mean for the audience here who don't know what the Creative Lab is? Well, you know how Google has the smartest scientists in the world trying to do amazing things?

6:38I like to think of Creative Lab as a place where I try to hire the best creatives and join the cause. And, you know, add the magic of human creativity to add a little bit of beauty, joy, meaning, humanity, and soul to the products that we ship. I love that. What's a bit of this soul that you've added to some Google campaign that we would know? Soul. Well, I think it originated way back when we did the Parisian Love, like literally a demo of someone using Google search that made people cry, which no one thought was possible. But all the way, even recently, the March Madness stuff for Gemini app hopefully made some people laugh.

7:22And they're also just demos of products, just showing people, because a lot of people actually don't know how to use AI and what they use it for. And we're just showing them the bunny soap, like, hey, you can do this, you can do this for this. But in a way that, you know, doesn't taste like eating broccoli. Although I love broccoli, no offense. This AI moment, for a lot of our customers, I speak to them, feels overwhelming. There is so much choice, there's so much to learn, everything is changing. So I'm kind of interested in the AI routine that two of you have. How have you reinvented and keeping ahead of just everything?

7:59And what can we learn from that? Well, maybe I'll start something. You can throw something in too. I try, so I do a lot. I do my best thinking often when I'm writing. And so every Monday, I've kind of made up this fake deadline for myself. I try to send out the most interesting things I'm seeing in AI to the team. And it's a very simple, like, I don't know, 10 or so bullets of things that are just kind of on my mind. Something that's like, huh, that's surprising. That catches your eye. And so that's one way I'm trying to just kind of put that together. And I don't outsource that to any agent or something.

8:34I actually do it manually. And I kind of think about the connections. That's kind of one of my weekly routines, I guess. So I get that email. And if you're inside of Google, you can just sign up for it. It's a great read. The funny thing is you can respond to Josh and say, hey, this is really interesting. And he responds back. It's like, I don't know how many thousands of people have signed up for this. But I really like the notion of taking the time for you to be inquisitive. And I think that's as marketers and as leaders, we have to get our skin in the game. We can't just rely on other people to serve it up for us.

9:04We have to do that work ourselves. Robert, how are you staying ahead of this AI wave? Yeah, I think three things. One, the very first thing about two years ago, when all this was coming at us really fast, I started an AI breakfast club at the Creative Lab, where literally once a week we just bring in some bagels and we say, hey, what have you read? What have you learned? And we're just sharing that stuff. That became a chat group within the Creative Lab. And now that is where everyone shares what they've learned or what prototype they've made. And it's interesting. that's like my first source of AI news pretty much.

9:43And because it's so much fun, it's intimidating to learn by yourself and it's so much fun to learn together and you're learning with a lot of people you respect. So that's really, really useful. I recommend anyone just start a AI chat in your working team. The second thing I do, and I've always said this, like just make sure you play with the stuff. And like you, you know, I try to try new things. You know, last week I made a music flow song for someone who was in the hospital. It's like just, you know, just play around with it and lower the stakes also. And I think the third thing is actually attitudinal.

10:15And that's something I try to do very much for myself and for the team is to lower the stakes because it feels pretty intense right now and the opportunity is insane, which is epic and awesome. But, you know, there's also a lot of fear in the system. And I heard Sean White recently, you know, the snowboarder, when asked, like, how does he prepare the U.S. Winter Olympic team? And he says for the young talent that he's trying to train, he just tries to, like, get him to chill and lower the stakes that when you go out for the run, even though it's for, like, a medal, it's just another run, another day.

10:55And I think that's really, really important. So I think it's really important to lower the stakes for your team to kind of say, like, hey, make them feel really safe because they, you know, there's a lot of fear, like, ooh, am I behind in AI? Will AI take my job? So all this stuff is in the system. You just lower the fear, make them feel safe, but at the same time, challenge them not to play it safe. And that sounds like contradictory, the two things, make them feel safe and challenge them to not play it safe. But actually, it's the same thing because when people feel safe, and creative people are some of the most insecure people on the planet, myself, they take bigger swings.

11:36And you want people to be relaxed to take the big swings. Can I just add something on that? I completely agree. And I will say, I wish all of you could spend just like an afternoon in the lab with Robert and the team. It's actually inspired a lot of how we tried to set up Google Labs too, is there's a looseness and a playfulness that comes with this. And I think you can, yeah, it gets stuck in the headlines and the back and forth and the jostling and all of these big giant things. And I think the really big unlock we've found is you get a small group of people together. You let them kind of dream big and you kind of just let them kind of create and jam.

12:12And that's been kind of the recipe for some of the things that have come out of labs. And it's weird because you can't microwave it or over-engineer it. It sort of organically just sort of happens when you get that chemistry and sort of that environment right. So we're going to relax and lower the stakes. I'd love to hear because, look, AI isn't perfect. We know that. It's getting exponentially better every week and every month and every year, but it's not perfect. So as you've been working through problems, what's the last time or recent time when you got really frustrated with what you can't do or it's just not working the way you wanted to do?

12:50And then how did you reset?

12:55Well, I had a few of these on the road to Google I.O. When a demo wasn't working properly, but we got it to work. I would say maybe one of the big lessons I've seen, Josh, over the last couple of years is there's been a tendency sometimes where we will, maybe when we're not doing as best as we could, where we'll kind of just take where the model is and just assume that's it. And we've had ideas that were maybe early that now work. But there's a few that I wish we gave up on too early. And so a good example of this is we had someone on the team, maybe two and a half years ago, was really passionate about, I want to be able, at the time, vibe coding wasn't even a term, but they had this vision of kind of software you could just create for teachers, like for educational kind of setting.

13:41And we tried that, and the models just weren't there. Not just at Google, just in the industry. And we kind of set that on the shelf, and now we've kind of come back around to it. But that was one where it was like, oh, we were just a little early. But to me, it was a really good lesson of trying to really push the frontier where it is and then kind of almost like have to reset. But we do hit those a lot sometimes. One of the things I think from a marketer's perspective is there's an expectation from the C-suite that it's all going to work. It's like Tom Cruise, a minority report. It's just magic is happening.

14:14And actually having some grace of, no, it's not perfect. But you try, you get knocked down, and you sort of get back up again, to quote Chumbawamba. Yeah, yeah, yeah. So far from perfect. I think there's a, you know, as a creative, you know, it's all about control. And you want the craft to control everything, every pixel. You want it exactly the way you want it. And AI is not great at that. It wasn't that great at that. It was terrible at that when early on. It's getting much, much better with Nano Banana and now Omni. Getting much better. But still not there. I mean, you know, when you have a pencil, you know exactly like your hand and that pencil and that lead and that piece of paper, you have complete control.

14:57It doesn't have that yet. So that's still frustrating. And I think but most more frustrating than that is people's expectation of AI. The like press a button. Come on, can you just press a button and the great things will happen? And the tough thing is like anyone who has that expectation has got to kill that expectation. And if you ever use AI, you realize, oh, shit, it's really, really hard to get it to do amazing things, especially in the craft high bar prep. Not so much the scale and efficient part. That part is really, really good at. Yeah. I would say, too, what we're seeing together when we do a product like Google Flow is we're trying to bring both that control as well as the ability where you can explore.

15:39And I think that's actually one of the greatest maybe potential things I see right now is as the control ability gets better, you can control the character or the scene consistency or other things. That's amazing. But you want to be able to keep it loose and light and keep the latency down where you can explore lots of different ideas. And I think that's what we see with a lot of our tools we're building right now, especially for creatives, is like how fast can you let people explore like 20 terrains at the same time? And then, of course, you can go deeper and sharpen those. But it's very much an explore and refine process.

16:15It is not a slot machine, hit the button, or pull the handle and sort of get what you want. 100%. Yeah. So the two of you worked on flow together. And I think there is a larger leadership lesson here for all the marketers in the room that in this AI era, the people we work with inside our companies are changing. The CTO, the CIO, technologists, and that's not always been the case for marketers. So I'd just love to hear from, what have the two of you learned from each other as you helped co-create Flow? And by help co-create, you built it and Robert gave you some advice. But far from that, maybe I want to start with that actually, because I think for, well, maybe a lot of you and your teams, Flow is a product and I'm just curious, maybe show of hands, has anyone used it or tried it a little bit?

17:03Okay, cool. I would say, talk about the soul of a product. It came a lot from Creative Lab and thinking about how do we build a tool that respects the creative, that respects the process, the craft. This is not a hit a button, automate all of creativity. It is more how do you bring that controllability, the ability to customize, the ability to get lots of different shots and inspiration. that was very much kind of like the heartbeat of what I learned from Robert and his team when we were working on it together. That was one. The second one is good products always make you feel something. And I think one of the awesome things, you're probably, you're seeing it on stage here too, but imagine Robert and a whole team of people that when you have a meeting with them, well, one, you leave the meeting feeling great, but two, that sort of seeps into the product.

17:58And so when you use Flow, we care a lot about the little details and the designs and the animations and the transitions and all that stuff. You're almost like a chef cooking for chefs, right? I mean, the people using this product, we want them to be the professionals that are really going to push it. And so those would be two things right off the top that very much kind of came through. That's very nice of you to say. I mean, I would basically all I want to do when I grow up is be Josh Woodward. No, I think the most inspiring thing about Josh when you work with him is eternal optimism and kindness and generosity.

18:37And the other thing is just obsessiveness about the user. And in this case, lucky enough, we are the users, so we are the creative. So he's obsessing about, what do you guys think? What do you guys want? What sucks? So it's really good. But like that obsession with making it great for the user and also thinking that like, you know, no limits, whatever you can think of, we'll figure out how to build it. You know, so it's very inspiring. Was there a moment for either of you where there was some friction, some tension, miscommunication, mistranslation? You are in very different disciplines or was it all just sweetness and light?

19:11Absolutely no conflict whatsoever. No, there was one moment where I had to like go, Josh, one of your PMs is like overriding some of our designers instincts. You remember that call? Oh, yeah. I'm like, you know, like this is for us. Remember? It's like, yeah, yeah, yeah. I'll have a talk. And next thing you know, the next meeting was better. Yeah, yeah. Well, I think one thing, too, I'll say there was a little bit of a process because remember the team building it, they are not. that they're creative with code, but they've never made anything like Robert's team. I mean, they're making Super Bowl ads and shorts and social, everything, basically.

19:49So they really are the experts. They're the ones that the product's built for, and hopefully all of you, too. And I think one other kind of lesson through that was, and it's been fun to watch the teams kind of co-evolve and co-create this way, but yesterday's announcement with Flow, one of the things I'm very excited about, which was actually kind of a co-developed idea, was this idea of tools that are going to be in flow. And you can look at that and say, oh, yeah, those are just plugins. I've got that and whatever else. But kind of the inspiration for this is that we think software of the future, not just creative software, but definitely creative software, is going to be molded and shaped by the people that use it.

20:26And in the past, when you created software, you, as the developer, had to come up with all the features, and it was one-size-fits-all, and it was very actually limiting and constraining, especially for creative audiences. And so why I'm so excited about Flow Tools is one part is we kind of did it together because the teams now are in sync. But the other part is I think it's one of the first times our software that we write is actually kind of a thin layer and you all are going to be able to kind of mold it and put in exactly what you want in the feature. So people are writing shaders and effects and text overlay, you know, all kinds of things.

21:04So it's sort of a blank canvas inside the product. That's the most exciting thing I think for creative people I think because before you have to learn the tool now the tool learns you that's right and With that one feature launch of Flow tools, you know, because you know, we used to get like thousands of feature requests or repeat making a thousand feature requests and like God damn it. They're not launching fast enough And then there's like you know you make it yourself and basically we've made the platform so anyone can make it themselves which is awesome. In natural language. You just describe what you want.

21:34All you need is words. Yeah. I think that point is so fundamental. And this is the transition from a marketing world where we just try and make the way we used to do marketing 10, 20, 30 % cheaper or more efficient to a world where you actually imagine what can I do with AI in a way that's never been done before, a whole reshift of how we work. It's super exciting. We've got a few minutes left. A lot's been announced. There's a lot inside of Google anyway. Robert, Josh, what would your advice be? Where should marketers here in this room and listening to the pod, what one or two or three things should they do to lean in, to coach themselves, train themselves?

22:15What Google products matter right now you think would be best? Go for it. Well, my go-to is, I'm not brown-nosing. He does run Gemini. I say Gemini because Gemini is the all-purpose tool. know, tool. It could be powerful. It could be bunny slope. You could make music, make video, make images. I think it's just really, really great. And I think you don't even know what it could do. Like, I'm always telling you, I wonder if it could do this. What? It does that. So what's something you've done recently with Gemini that was going to either blew your mind and you thought, this is going to make me better at my job?

22:48Well, okay, it's not a job-related thing, but literally at lunch, I was playing around with Spark, the agent that was just announced. I'm looking to real estate, potentially downsizing. My kids are growing up. And I just made an agent to like scrape all the real estate things with a very, I have a very specific criteria, low maintenance costs as they go into, think about retirement someday in the far future. and I told it to keep me and alert me when the new things come up. And it's like it made a dog. It laid out the – it's amazing. So just imagine doing that for all parts of your workflow. Okay, so all the marketers should think about buying a new house and using Spark.

23:35I get it. Josh, what Google products – what should people lean into? Yeah, I mean, I think before we get to any Google products, I think the thing I would be trying to think a little bit about is for each of these different mediums, whether it's images or video or audio, Google has a lot of stuff to offer in each. And I would be trying to think about how can you kind of play with what's the best way to get a feel for image generation and editing or video or audio. Because I think for all of you, you're working across many channels and many geographies and everything. So I think that's just as a general thing, And the pace of this is changing so fast.

24:13And I hope all of you will get a chance to play with Gemini Omni. And the ability to just edit video, just like you maybe have done with, say, editing images in Gemini, it just blows your mind. There's one thing, just real quick as a quick side. We did a podcast yesterday. And imagine this podcast just recorded here, like, for your show. And people just started superimposing, like, you're holding a plant. Robert's petting a cat. A parrot lands on my shoulder. But all this stuff is now possible by just describing it. And of course, that's kind of gimmicky and whatnot. But you can imagine the ability to kind of recast scenes and characters and other things.