Adobe opens public beta for Firefly AI Assistant with 60+ tools across Creative Cloud apps

Adobe has launched public beta access to Firefly AI Assistant, an AI agent that orchestrates multi-step workflows across its Creative Cloud suite from a single conversational interface. The assistant coordinates actions across more than 60 tools spanning Photoshop, Premiere, Express, Lightroom, and Illustrator.

How it works

Firefly AI Assistant allows users to issue natural language prompts that trigger coordinated actions across multiple Adobe applications without requiring detailed knowledge of each platform's specific tools. The system maintains context across sessions and can execute complex, multi-step creative workflows.

Adobe has introduced "Creative Skills" — pre-built agentic workflows covering common tasks including batch photo editing, mood board creation, portrait retouching, and social asset generation. According to Adobe, the assistant learns user preferences over time, including preferred tools, workflows, and aesthetic choices.

Users retain full control throughout the process. They can interrupt at any point to manually edit, adjust layouts, or modify assets using traditional tools. The assistant includes context-aware features that can create or use adjustment tools on the fly based on the specific artwork being processed.

Availability and pricing

The public beta rolled out globally on April 27, 2026 for Creative Cloud Pro subscribers and users on paid Firefly plans (Pro, Pro Plus, and Premium). Pricing for these plans was not disclosed in the announcement.

Adobe first announced Firefly AI Assistant in late March 2026, promising a "soon" launch for the public beta.

What this means

This marks Adobe's entry into AI agent territory, moving beyond single-app AI assistants to cross-application orchestration. The 60+ tool integration suggests significant engineering effort to create APIs and coordination layers across Adobe's ecosystem. If execution matches promise, this could reduce the technical barrier for complex creative workflows — though actual performance in beta will determine whether it delivers on that potential. The success will hinge on how reliably the agent interprets user intent and whether the pre-built Creative Skills cover enough real-world use cases.