Inside Adobe's AI Strategy with CTO Ely Greenfield

0:07Corey:Hello everyone and welcome to the Neuron. I'm Corey Knowles joined as always by our trusty writer Grant Harvey. Hello. We're coming to you today from Adobe Max 2025 in beautiful downtown Los Angeles and we have an extra special guest. We're joined today by Eli Greenfield. He is the Chief Technology Officer and Senior VP of Adobe Creative Products which puts him on a list of the latest features. top of the entire creative suite and we're here to talk today about some of the great things they've had going and announced this week. Eli, welcome to Neuron.

0:37Ely Greenfield:Thank you. Very excited to be here as a longtime subscriber. Excited to be here talking to you guys.

0:42Corey:That's awesome. Well you know I noticed, I guess to start off, there was a real push it seems like this week toward practical AI and this idea that of maybe consolidation and interconnectedness a little bit between various tools, specifically with Firefly. How are you thinking of this kind of reimagination from a technological standpoint?

1:04Ely Greenfield:Yeah, it's funny. I think you're right that, you know, with what we showed on stage and what we announced this week, it really, you see that connection of all this coming together practically. For us, it's not a new push. I mean, it really is, you know, since we really turned our investments towards focusing on this new wave of Gen AI technology, it has always for us been about for our customers, you know, for the creative industry, how do we take these amazing tech demos that everybody was seeing and make it actually something practical that they can use in their hands? You know, the ability to type a prompt and generate an image or a video and have it appear out of thin air is amazing.

1:43Ely Greenfield:But for our customers, you know, what they need to figure out is how do they put that in their workflow and how they make sure they bring their creativity and their craft and their expertise to that. And it becomes another tool in the tool chest. So that's what our investment has been all along. But yeah, I think you're right that with what we showed and announced in the past few days, it really is turning a corner where the independent investments that we've made in AI are starting to work their way up into the rest of the platform and the tools and really starting to show up in workflows and use cases that make our customers go, I see how I can put that to work.

2:22Corey:One of my favorite things yesterday was everybody was so excited about all the announcements, but the one that everyone just lost it over was when they talked into a chat agent and said, rename these labels for me. These layers for me.

2:34Ely Greenfield:You know, what's great about this conference, we come to Max every year and we've been doing it for, I don't know, 25 years, something like that. We've always been showing off like mind-blowing technology you can find over the decades videos of max and people talk about like adobe black magic and we have amazing researchers and we're always very proud of that work but we come to max and we show it and every year there's one feature like that which is in some ways the most boring banal application of technology right that the crowd goes nuts over because you know these are people who live and breathe in these tools all day long and the creative piece that's what they love and And the boring management pieces, that's the stuff that, you know, it's a major time suck for them.

3:17Ely Greenfield:And so it's a good grounding for us to remember application of this stuff to real world problems and to just the grunt work of applying, you know, the stuff in between the creative pieces. That's how we can serve our community most. So it's great to hear that applause and be like, oh, right. Okay, we've got to do more of that stuff.

3:34Corey:That's right.

3:35Grant:And there was a lot of those. I was particularly surprised by how practical and hands-on a lot of the tools were, where it's like, oh, I can see this being applied in creators' workflows every day. Because that was just one of them. There was also Harmonize, which seems like that was a huge one. And then also, I don't know how long you've been doing this, but the one where now all of a sudden you can automatically map in Premiere the individual skater and the skateboard as two different objects. and just no hours of rotoscoping. It's already in there.

4:08Ely Greenfield:Yeah, it's funny. If you look at the history of development in all of our flagship tools, Photoshop, Illustrator, Premiere, After Effects, all of them, we have this well-understood core set of tasks to be done for these people and for our customers. And things like being able to accurately select something or being able to composite two pieces of an image. Photoshop is a compositing tool. So how do you bring two things together and make them feel like they belong together? Those are their new wizard like technology we're constantly innovating on. But that idea of, you know, correctly select and mask an object like that's that's been a core problem statement for our customers for decades.

4:51Ely Greenfield:Right. Harmonizing two things. That is a core problem statement. And in many ways, again, the research and the technology we put out is always how do we improve that practice and, you know, reduce the time, make it more productive. Gen.ai, you know, I kind of like to talk about it as for 35 years now, you know, the first version of Photoshop, right, sort of invented digital art, digital images. And it was black and white. But other than that, you know, like any image you can make in Photoshop today, you can make back then with a pencil, right, assuming it's black and white. But back then, it was literally flipping individual bits on.

5:26Ely Greenfield:And everything we've done since then is, how do we take creative productivity-enhancing technology to sort of elevate that conversation so that the human can, again, invest in their creative work, think about what they want to do, bring their expertise to the table. But the production aspect of it, it's more and more of asking this assistant to realize their vision. And this is kind of another step in that. So all those examples you gave, I mean, I think the latest ways of AI give us more opportunities to improve that. And again, help people save more time doing that. But yes, in some ways, the problem statements are not new.

6:00Grant:That's also true with the agentic workflows that you've basically integrated into Express and Photoshop. Like the Express demo was really impressive where basically you're chatting with the assistant, having the assistant do a lot of the editing for you. And then you can take over at any time and say like, hey, I actually know I can do this faster than you. or that's not exactly what I want. So let me just get in there, tweak it. And it's down to the layer level. Yeah. It was very impressive to the point where you can even generate custom UI elements, right? Yeah. That's really cool too. How do you do that?

6:32Ely Greenfield:Magic, black magic. I mean, there's a really interesting analogy that I keep kind of discussing with people. If you look at what's going on in the vibe coding space, right, or I'll say in the AI assisted coding space, there's kind of two ends of that spectrum. There's the pure vibe coding side, which is, hey, you don't have to look at the code. You don't have to think about the code. You're not a coder. You don't care how it's implemented. You just want to be able to talk with the assistant and you get results. You get an app out of it and it's great. Now, if you don't like what you get, you're kind of out of luck.

7:04Ely Greenfield:You can kind of coax the AI the direction you want, but you're not going to get in there and mess with the code yourself versus the other side of it, which is you're in an IDE, you're writing code, you've got the assistant there. and it's not changing the tools and the structure of what you're working with. You're working with code and projects, but it is there to help you. But it is really a collaborator that you can hand tasks off to it. You can take them back from it. You can dive in and use all of your expertise, and it becomes a partnership or an assistant you hand off to. So I think when you look at the media space, as we look more and more at Agentic, we're probably going to see those same two poles.

7:43Ely Greenfield:And it's obviously a continuum, But there are people for whom they don't have the production skills. They don't know how to use our tools or other people's tools. And they may have creativity. They may have an idea of what they want. But they're willing to hand it all off to an agent to say, you figure out how to do it. I don't need to know anything about how you do it. Versus those who will say, if you can do this, great. But if not, I'm going to take the reins. And so, again, for a lot of our customers in our flagship tools, that's really what they need is that idea of, great. If you can help automate the SCUT work, the repetitive production work, cool, but don't take control away from me in the process.

8:21Ely Greenfield:Make it additive. We have plenty of customers for whom that other side is important, too, and we're investing in those as well. But for the creative professionals out there, it's that add, but don't subtract. That's kind of the philosophy.

8:35Corey:Yeah, because there are a lot of people specifically in the artistic space, above all spaces, I would say even, who are more resistant to AI and their approach and in things they're touching. But I think, I thought it was really interesting and really telling of your understanding of your audience and of the people who use your tools that this was very much presented as a let us help you with the mind-numbing awful work. let us help speed you through the simple stuff and allow you frankly more time to be creative

9:05Ely Greenfield:with any luck the creative industry you know a lot of people that have had this reaction which is whoa i don't want to go near that stuff and i think i think part of the reason of that is because the first you know earliest iterations and for a while the the technology that has been presented has been subtractive to them it has been type a prompt get an image awesome what if i don't like it you can roll the dice again but you know there there's there's a graph that i like to draw and you know if we had a mark on whiteboard i could do it but you know if you if you kind of try and graph the the time and energy investment you spend in creating a piece of content on the x-axis and the y-axis is is the value of the result or the you know quality of the result um you can plot different tools on that and they all have different curves and some of them you have put a lot on time you know, but you can create anything.

9:53Ely Greenfield:Some of them kind of start to top out. I realized, you know, describing a graph is great podcast material, but I'm going to run with it. I'm going to push through.

10:01Grant:AI crawlers, what are we going to do about them? To let them crawl or not to let them crawl? Well, if you are running a small or medium-sized business, you really need to see what other SMB leaders are doing about this. Luckily, Clutch just released this comprehensive new report breaking down exactly how small businesses are navigating AI crawlers, and the results might surprise you. See, Clutch surveyed 320 SMBs to get their reactions, finding 72 % of businesses are worried about them, 71 % have been hit by bot traffic already, and 42 % even cite bandwidth issues coming from AI. But on the flip side, 62 % say AI search has been a net positive, with 60 % reporting more revenue, 42 % more traffic, and 44 % more leads.

10:49Grant:However, 57 % are still just blocking AI crawlers entirely. So digital pests or secret growth hack? Either way, crawlers are here to stay. Just make sure they're working for you, not against you. To dive into the full clutch report, just click the link in the description or go to clutch.co and search AI crawlers. Again, Then click the link below or go to clutch.co and search AI crawlers.

11:15Ely Greenfield:The first generation of generative AI media technology, the line kind of looks like this, right? Like you type, you literally spend two seconds on it. You get something pretty good, but there's very little you can do to improve that. And that's what drives the creative community crazy. It's like, wait, I have all this ability, but I can't get better than what anybody else can get. And so you're, you're basically trying to replace me with this thing. So that's just not helpful to them. So our whole focus is how do we put a graph in that line that starts them up nice and high, but gives them the ability to shoot to the moon way faster than they could have with previous generations of tools.

11:49Ely Greenfield:And knowing where their destination is because they've got that creative ability, because they've got that expertise on what content do you want to achieve this goal.

11:57Grant:With control at the layer level, which to me has always been the flaw of generative AI, is that like, okay, this is great that I can make an image out of nowhere, but I can't manipulate this person's hand and make it go here and there. I watched you.

12:09Corey:And the resolution isn't there.

12:10Grant:Yeah, the resolution's not there. So the generative upscaling that's available with Topaz Labs now in Photoshop. That's so cool.

12:18Ely Greenfield:That is one of those things that a lot of people are using it already. So having it in there, it's just solving the problem that they have.

12:25Corey:I keep joking that that is like, you know, I've been watching police procedurals for many, many years where they're like, enhance that image. And they zoom in and they zoom in and they zoom in. and you've got this thing that suddenly is very clear. And I'm like, this is not how images work. That is not real at all. It is now. But it is now. Yeah, exactly.

12:44Ely Greenfield:Whatever your domain or your industry is, as soon as somebody puts it on TV, you get immediately frustrated because they're like, that's not real. And those of us who've worked in technology and media have been seeing those things for decades. They're like, oh, God. And then all of a sudden, somebody's like, no, you're wrong. Actually, it's totally doable. And you're like, oh, my God.

13:01Corey:That was my first reaction yesterday. day. You know, something else I think you guys called out really well and that you're doing that really sets Adobe apart and in my opinion shows that you understand your place in the design and artistic industries as a whole is you promised three things yesterday. You promised to continue to develop your own models. You promised to continue to explore good opportunities for partner models to have available in the tool. And you also promised the ability to do your own models and train your own tools. And I'd love to learn a little more about the thought process that went into that because that feels like the right approach.

13:40Ely Greenfield:Yeah. So, I mean, that's been a journey we've been on. You know, I'll say when we first, again, rolling back to when we first started thinking about how do we bring Genitive AI in here, we looked at the models that were out in the world. We looked at other providers and we talked to our customers. And what we heard at first was basically none of our customers thought you know any of the stuff that was out there was something that they could use and yeah number one reason is for the reason that we've been talking about is control is okay it doesn't it's just not something that we can integrate into workflows well yeah and number two was you know the the early models and the first injection of AI there are all these questions about the legality of it the ethics of it the you know what does it mean who owns the output you know what what happens and what it's trained on.

14:27Ely Greenfield:And so we looked at both of those and we said, okay, there's a gap in the market here that's not filled, but there's so much value we could provide if we can solve those two problems. We need to fill that gap. So we started building models to do that. At the end of the day, right, we don't sell models, we sell tools, we sell, you know, the ability to realize your ideas, you know, sort of the technology that helps you create the content. And so that's always been our driving factor is okay what what are the things that our customers are going to need how can we what kind of new capabilities or workflows can we provide for them that meets their requirements if there's something in the market that's going to meet that that's great yeah love to provide those and we want to give people the choice of using those yeah and if there's not i have an amazing team of some of the world's best scientists ai experts computer vision experts who can then create that future that we want to have happen.

15:20Ely Greenfield:And so that's always been our philosophy. And you're just seeing the market develop to the point where, and the technology develop to the point where, great, we can start to offer all those solutions. So we will absolutely continue to build our own models because that gap in the market still exists. And as we imagine new ways that we can empower our customers and give them ways to create using technology. We will continue to find those gaps and fill them. Our rocket scientists will continue to figure out how to create that technology. But there are amazing models out there, and we know that, and we hear it, and we're very close with all those other companies building them.

16:01Ely Greenfield:And so the more that we can give customers the power of the technology merged with the workflows and the tools that these people, our customers live and breathe in, like that is, that's a win for everybody. And then that third one is that build your own models, right? And this is, we've had this technology for a little while, you know, as we've been developing, we started with just, you know, a small number of customers that we worked with, but have now brought it to the point where we can really make it broadly available. And as we showed this on stage yesterday, you know, one of our evangelists,

16:32Corey:who's just a fantastic illustrator, Paul Trani, you know, he has all this amazing work he's done

16:38Ely Greenfield:over the years, developing his own style, his own look, themes to content. And so again, it is absolutely his unique work and his creativity. But if we can give him the tools to take that, say, let me train a model on this. And now I can take that concept I've developed and really run with it and see, can I use the AI to help me apply this to new layouts, new concepts, new content, new situations, new objects, and really explore the idea further. Again, it's a massive production productivity enhancing tool for him. And it makes his creativity go that much further. So very excited to make that available to everybody as well.

17:21Grant:And he can create multiple different iterations. So if you're a brand and you have multiple different styles, or you're a contractor, you work for multiple different companies, you can create kind of seemingly infinite different models for your different clients.

17:36Ely Greenfield:Exactly right. And if you go, you know, follow any artist and you'll see them do this, they explore, they develop a style, and then they exploit that style. You know, it sounds like a dirty word, but it's not meant that way. No, explore and exploit.