Adobe introduces the Firefly AI Assistant to manage tasks throughout Creative Cloud.

      Adobe has introduced the Firefly AI Assistant, a conversational tool that coordinates tasks across Photoshop, Premiere Pro, Lightroom, Illustrator, Express, and Frame.io through natural language commands. Formerly referred to as Project Moonlight, it will enter public beta in the upcoming weeks, incorporating third-party models like Anthropic's Claude and retaining context throughout different sessions. Additionally, Adobe announced Firefly Image Model 5, Custom Models, and the Project Graph workflow system, which utilizes nodes. This release comes as CEO Narayen is set to step down, with Adobe facing challenges from competitors such as Canva (with 260 million monthly active users) and Figma.

      The Firefly AI Assistant functions as a conversational agent capable of executing creative tasks via plain language across several Adobe applications, eliminating the need for users to switch between tools and navigate various menus. For example, a designer can simply instruct the assistant to resize images for social media, color-grade footage according to brand guidelines, or generate multiple logo variations, and the system will manage the necessary processes across the relevant Adobe software.

      Entering public beta soon, the assistant was first showcased as “Project Moonlight” at Adobe MAX in October 2025. It can remember project parameters, brand guidelines, and past decisions between sessions, rather than starting from scratch each time. It also integrates with Frame.io for smooth feedback and approval workflows directly into the assistant’s task management.

      Adobe confirmed that the assistant will interface with third-party AI models, including Anthropic's Claude, alongside Adobe’s Firefly models and those from partners like Google, OpenAI, Runway, Luma AI, ElevenLabs, among others. The new image and video editing features introduced with the assistant are immediately available for Firefly plan customers.

      The Firefly AI Assistant signifies Adobe's entry into the era of agentic software, moving away from traditional tools that respond to isolated commands towards systems that comprehend intent and autonomously execute complex workflows. This marks a notable shift for a company built on individual applications, each with unique interfaces and subscription models.

      This strategic shift comes amid competitive pressures reflected in Adobe’s financial reports. In fiscal year 2025, Adobe generated revenue of $23.77 billion, with annual recurring revenue from digital media at $19.20 billion, showing an 11.5 percent year-on-year increase. The goal is to reach $26 billion in revenue for FY2026. Although these figures are impressive, growth rates are slowing, and Adobe’s stock has dropped approximately 43 percent amid investor concerns regarding the viability of its traditional application model in an AI-driven market, where competitors are offering increasingly sophisticated tools at lower prices.

      Canva now boasts over 260 million monthly active users, primarily targeting small businesses and marketing teams similar to Adobe's Express product. Figma holds about 80 to 90 percent market share in UI and UX design—an area Adobe attempted to acquire for $20 billion before regulatory intervention. Both Canva and Figma are developing their own AI-driven creative assistants, without the historical baggage of legacy product suites predating large language models.

      Essentially, the Firefly AI Assistant acts as an orchestration layer above Adobe’s applications. Users can instruct it to take a raw photograph from Lightroom, apply specific editing styles, create three differently sized versions in Photoshop, generate social media graphics in Express, and prepare everything for review in Frame.io, all through a single conversation. The assistant determines which application to use for each task and executes the necessary steps.

      This matters significantly, as multi-application workflows often lead creative professionals to waste time. For instance, a video editor using Premiere Pro may need to manually coordinate the design of a title card in Illustrator, color-correct footage in Lightroom, and process audio in Adobe's tools. The assistant aims to streamline these tasks into a unified interface, making the applications themselves more transparent and focused on results.

      Additionally, Adobe will be launching Firefly Image Model 5 in public beta, which is its latest image generation model, and expanding Firefly Custom Models that allow users to train a model on their specific image libraries for consistent visual styles, character design, or photography aesthetics. Custom models are private by default and reusable for multiple projects, catering to enterprise teams and studios focused on brand consistency without risking the exposure of their assets to third-party data sets.

      Alongside the assistant, Adobe is also working on Project Graph, a visual workflow system that allows creators to design, link, and automate AI-powered workflows across Creative Cloud. While the Firefly AI Assistant relies on natural language, Project Graph provides a visual interface for users to connect AI models, Adobe's tools, and effects into reusable "capsules,” which are portable templates that can be shared across teams and utilized within individual applications.

      Currently in development, Project Graph demonstrates Adobe’s long-term strategy that the value of its platform lies not in any one application but in the integration between them.

      This push into AI comes as Adobe experiences a leadership transition. CEO Shantanu Naray