You’re reading an issue of “The AI Economy,” my newsletter exploring the forces shaping the AI era—tracking how AI is rewriting business, work, technology, and culture. Subscribe to get expert insights and curated updates delivered straight to your inbox.
For decades, learning Adobe meant learning its tools — the shortcuts, the workflows, the muscle memory of navigating a suite built around precision controls. On Wednesday, the company argued that shouldn’t be a prerequisite anymore, unveiling Firefly AI Assistant, a creative agent that consolidates the power of Photoshop, Premiere, Lightroom, Illustrator, Express, and more, all within a single conversational interface.
If this sounds familiar, it should. Firefly AI Assistant is the evolution of Project Moonlight, the initiative Adobe previewed at MAX last October and has since been advancing through private beta. Its arrival as a commercial product marks a core addition to the company’s product roadmap, signaling that agentic creativity is no longer something Adobe is toying around with. Firefly has become the connective tissue of its entire creative platform, in the same way Slack became the unified interface through which Salesforce’s workflows now run. Adobe is making the same play, just for creativity.
The assistant doesn’t arrive alone. Adobe is pairing the launch with new video and image editing capabilities in Firefly and the addition of Kling 3.0 and Kling 3.0 Omni to a partner model library that now spans more than 30 creative AI models, all unveiled as part of the company’s annual AI Summit.
“Adobe is leading the shift into a new era of agentic creativity, where you direct how your work takes shape and your perspective, voice, and taste become the most powerful creative instrument of all,” David Wadhwani, Adobe’s president of its creativity and productivity business, says in a statement. “Adobe Firefly is a category of one, with the best models, the most powerful tools, and now, a fundamentally new way of creating that gives you the combined power and precision of all our creative apps in one place.”
From Tools to Outcomes
When Adobe released Firefly in 2023, it started off as a standalone generative AI tool—a place to experiment with text-to-image generation kept deliberately separate from its creative suite. The company was treading carefully: its core customers were the professional creatives most threatened by generative AI, and Adobe wasn’t ready to make any promises about where this was all heading.
Over time, Adobe began embedding Firefly’s capabilities directly into its core apps, including Photoshop, Illustrator, and Premiere, shifting it from a destination into a capability layer woven throughout the suite. But the starting point for any creative project was still the tools. Were you making a video or editing a photo? That determined where you began. Adobe believes it’s time for a change.
“Adobe’s been in the business for over 40 years, building incredible professional tools, and those tools have buttons and controls, and your creative vision is kind of mapped into that environment. And over time, people have gotten very, very good at working from the creative vision through the tools to the outcomes,” Forest Key, Adobe’s vice president of agentic innovation for Firefly, states in a press briefing this week. “What’s exciting about agentic creativity is that you’re able to kind of think about the end game, the outcome you start at the end, and this is done in a semantic context. You’re using your language to express yourself and thinking about what it is that you want to accomplish.”
Central to Firefly AI Assistant is a fundamental rethinking of how language maps to the creative process. Key argues that every stage of creative work—from ideation and organization to scaling and sharing—can be expressed semantically, through natural language, with the agent handling the translation into tools and generative outputs behind the scenes. It’s a shift he describes as moving from mapping a creative vision through tools to starting at the outcome itself.
“These words are the creative process,” Key says. As someone who has spent four decades in the creative industry, he notes that the implications are significant. “Imagine being able to define what it is you want to accomplish, and then in a chat, conversational flow, being able to work with the agent to finish the project,” he remarks.
Now, instead of opening up a Creative Cloud app and working within the constraints of that program, you can let Firefly handle the execution for you. Just converse with the Firefly AI Assistant as you would with Claude or ChatGPT, describe what you want created, and Adobe’s agent will handle the rest. It will orchestrate and execute complex, multi-step workflows across Photoshop, Firefly, Premiere, Express, Illustrator, and Lightroom. Everything is done within one conversational interface—no context switching required.
Firefly AI Assistant isn’t operating autonomously. Adobe has been deliberate about keeping creators in the loop, allowing them to step in, redirect, and refine at any stage of the workflow. Nothing changes here. The assistant executes; the creative vision remains yours.
Adobe ships the assistant with a growing library of prebuilt Creative Skills, purpose-built shortcuts that handle common multi-step tasks like auto toning, vectorization, font search, style presets, and animation without requiring you to manually sequence each step. Beyond prebuilt skills, the assistant also introduces a new class of context-aware, natural-language-driven editing controls that sit alongside traditional precision tools rather than replacing them. Instead of a generic slider, you might see one that adjusts the amount of ice in a coffee drink or the density of trees in a forest scene, for example.
The assistant isn’t limited to just the Firefly app, either. As you switch to other Adobe programs, such as Photoshop, the agent will follow you, maintaining the same context in the same conversational interface you know, so you can continue building without having to explain things all over again.
Over time, it will learn your preferences, remembering your most-used tools, workflows, and aesthetic choices. Think of it as your own Nigel Kipling, the trusted creative right hand to Miranda Priestly in The Devil Wears Prada—someone who knows your aesthetic instincts, anticipates what you need, and gets it done without requiring you to explain yourself every time.
For decades, the central creative question for Adobe users was never really “what do I want to make?”—it was “how do I build it?” Mastering the tools, sequencing the workflows, navigating the suite—that technical fluency was the price of admission. Firefly AI Assistant is Adobe’s argument that the price has been waived. With the assistant handling orchestration and execution, the creative process can finally start where it was always supposed to: with the idea itself.
That said, this isn’t Adobe’s first foray into conversational creativity. In December, the company announced that Photoshop, Express, and Acrobat would natively be supported in ChatGPT. This meant that anyone could start designing for free using these popular creative apps, all from within one of today’s most well-known chatbots.
Adobe’s Firefly AI Assistant is expected to start rolling out in public beta later this month. The company says it plans to extend the assistant’s agentic capabilities to third-party chat platforms, including Anthropic’s Claude, enabling creators to access Adobe’s creative tools from surfaces outside the company’s own ecosystem.
Firefly’s Got More
In addition to the AI Assistant, Adobe is launching new Firefly updates, starting with video editing. They include the following:
- Support for “Enhance Speech”: This capability was initially found in Premiere and Adobe Podcast. When used, it cleans up dialogue.
- Better color adjustments: Now, you can fine-tune exposure, contrast, saturation, temperature, and other visual elements in Firefly Video Editor using intuitive sliders.
- Support for Adobe Stock files: Looking for that high-quality video, image, audio file, or sound effect to include? Look no further because Adobe Stock is directly integrated into Firefly Video Editor.
There are also two new image editing tools being released today:
- Precision Flow: Using a single prompt, Firefly will refine images by generating a wide range of results. For example, if you wanted to change the background of your image from a snowy setting to a rainy one or a desert climate.
- AI Markup: This provides you with more granular control over where and how edits are applied to your image. You can use a brush, a selection tool, or reference specific areas of the image to refine.
Lastly, Adobe continues adding third-party models to its collection, building the creative equivalent of what Microsoft Azure AI Foundry and AWS Bedrock have done for enterprise AI—a single platform where creators can access the industry’s best models without having to shop around. Admission isn’t unconditional, though. Partner model makers must agree not to train on creators’ work, and all generated assets must carry a digital signature and verified certificate identifying the model used to create them—a safeguard Adobe has positioned as central to its commercially safe approach.
The latest additions come from Kling. Its 3.0 and 3.0 Omni video models can now be used within Firefly. The former is optimized for fast, high-quality production with smart storyboarding and audiovisual sync, while the latter adds advanced control and consistency, allowing you to define shot duration, camera angle, and character movement across multi-shot sequences.
Kling 3.0 and Kling 3.0 Omni join the ranks of more than 30 AI models that can be used in Firefly, including Google’s Nano Banana 2 and Veo 3.1, Runway Gen-4.5, Luma AI’s Ray3.14, Black Forest Labs’ FLUX.2 Pro, ElevenLabs’ Multilingual v2, Topaz Lab’s Topaz Astra.
All of these capabilities and models are now available within Firefly.
Featured Image: Credit: Adobe/screenshot
