1-on-1 Interview With Adobe's David Wadhwani: Adobe Max 2025! | Frank McShan

I'm doing great. Thank you for joining us.

Crazy as always. I started coming to Max 22 years ago. Every Max has its own vibe and moment, but this one, the amount of innovation combined with the energy in the crowd has been off the charts.

It's the largest creativity conference in the world. We do it every year, typically in LA, sometimes in Miami. We gather over 10,000 creative professionals, including creative pros, creators, influencers, and marketing studio folks, all in one place to talk about creativity. We discuss tools, workflows, and innovation, and they share what they're doing with our tools. It's a great place to exchange ideas. It's super cool and I'm so happy to be here this week.

I can't pick one. We had so much innovation this year, we couldn't get it all on stage. We had the longest keynote ever and we were like, at some point we need to cut it otherwise we have to hand out diapers at the door. So we put some stuff in 'more coming.' The two things that really stood out for me: one is Firefly, the application. You saw this morning with Brandon B doing vibe creation on the fly. It's amazing how much you can do by blending direct manipulation of our tools with AI generation, and how fast you can do it. The Firefly application end-to-end feels amazing. The other one, you mentioned, is the AI assistant in Express. I remember sitting down with it, opening Express, seeing some content, typing and asking it to do things. It understood the context and made complicated changes for me, but also didn't get in my way so I could click and drag if I wanted. Those two things really stood out.

First and foremost, our approach to conversational experiences, which is what ChatGPT is, is different from what everyone else is taking. Everyone sees conversational as a mechanism for a straight linear conversation. We think, especially for creative expression, it has to be a combination of creation and direct manipulation. So we started building that into Express and Photoshop. When ChatGPT came out with its new developer SDK for the first time, they allowed us to integrate UI elements into the conversation. That was a big unlock for us. For your audience, sometimes you want to have a conversation and ask the LLM to do something, sometimes you want to do it yourself. And when you ask the AI assistant to do something, it should give you a UI component back. For example, if you ask it to blur the background, you don't want to tell it blur 70% or 30%. It'll give you a slider, and you can tell it how much, then continue with the flow. That's the most important part of getting this right.