Greetings from Los Angeles. I’m here at Adobe Max, where Adobe is showing how quickly creative tools are evolving into full AI systems. This year’s event focuses on the convergence of generative media, automation, and agents that create alongside people.
Adobe introduced new Firefly AI models and announced that every Creative Cloud application, from Photoshop to Premiere Pro, will eventually include an AI assistant. These assistants can edit projects with descriptive prompts, generate custom voiceovers and soundtracks, and coordinate tasks across multiple apps. In Adobe Express, you can describe what you want changed, such as “make the raccoon wear a vampire costume holding a pumpkin,” and the AI adjusts layout, color, and composition automatically. In Premiere Pro, the new Generate Soundtrack and Generate Speech tools build music and narration that synchronize with video.
The next phase of creative work is agentic. Project Moonlight, still in development, functions like a creative director for social media campaigns. It connects assets across applications, learns your tone and visual style, and can post to several platforms. Firefly’s multimodal foundation models now extend beyond images to video, audio, and text, giving creators one system that understands the entire production process.
Adobe’s challenge is control. Professionals want speed and consistency while maintaining ownership of their craft. The new model-agnostic Photoshop, which supports third-party AI models such as Google Gemini 2.5 Flash and Black Forest’s Flux.1, shows how Adobe plans to do this. Users can choose the model that fits their workflow and keep every asset brand-safe, legally clear, and ready for commercial use.
The trend toward a highly agentic future is clear. I’m excited to see where this goes. Are you?
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.