Adobe introduced new AI-based chat assistants for its core products, including Express, Photoshop, and Firefly. The web version of Adobe Express now features an “AI Assistant” mode, allowing users to create or edit designs using text prompts in a conversational format. Users can describe desired changes to a project in plain language, such as “add an autumn theme” or “make it brighter,” and the assistant selects appropriate fonts, elements, backgrounds, or generates new images based on the request through Firefly.
Photoshop now features an assistant currently in private beta. It is located in the sidebar and helps automatically select objects, create masks, and perform repetitive tasks such as background removal or color changes. Users can consult the assistant for step-by-step help or delegate routine actions, significantly speeding up image processing.
Adobe has also added support for third-party models in Photoshop and Firefly. Users can now choose models like Google Gemini 2.5 Flash, Black Forest Labs FLUX.1 Kontext, and others for image generation and editing. All models are available through a single interface, and Creative Cloud Pro and Firefly subscribers can generate unlimited images until December 1.
Firefly has received a public beta version of the Image Model 5, which creates high-quality detailed images at a resolution of 4 megapixels. New features include “Prompt to Edit” for describing image changes in plain language, as well as “Generate Soundtrack” and “Text to Speech” for creating music and voiceovers for videos through AI. Premiere Pro has added “AI Object Mask” for automatically selecting people or objects in videos, and Lightroom features “Assisted Culling” for quickly selecting the best photos.
Adobe also announced an experimental project “Project Moonlight,” which aims to unify assistants from different products and consider user style through social media connections. The company is also working on integrating Express with ChatGPT, allowing users to create designs directly in ChatGPT via OpenAI’s API. New features are available in public and private beta versions, with plans to expand support for third-party models and platforms soon.
