Adobe unveiled a built-in AI agent for Photoshop at the Adobe Max event in London that will help perform complex tasks and automate multi-step processes. Users just need to type a query into a text field, and the agent will either perform the task or explain step-by-step what needs to be done. And Firefly can now use models from Google and OpenAI.
Image source: Adobe
The new agent is available in the updated Actions panel and works like a chatbot. Adobe demonstrated how the AI applies effects to text and backgrounds, saving users time. According to the company, the tool can handle almost any task that would normally require manual intervention.
At the same time, Adobe introduced a new version of the generative service Firefly, which unites tools for creating images, video, audio and vector graphics. The updated web interface is already available, and mobile applications for Android and iOS will be released later. Firefly is now based on improved models – Firefly Image Model 4 and 4 Ultra, providing detail in 2K resolution (2560 x 1440 pixels). The service is integrated with Photoshop, Premiere Pro and Express, helping at all stages of work – from the idea to the final render.
Image source: Adobe
For the first time, Adobe has included third-party AI models in Firefly, including Google’s Imagen 3 and Veo 2, as well as image generation via ChatGPT. The company emphasized that user data will not be used to train these algorithms, thanks to special agreements with partners. In addition, all of Adobe’s own AI models are trained on legal content, so their output is safe for commercial use. This is especially important for designers and photographers who want to avoid copyright issues.
The AI agent for Photoshop and a new version of Firefly are scheduled to be released in the coming months. The exact date and subscription terms have not yet been specified.