Adobe previewed a local system that would let third-party AI agents work on projects in four Creative Cloud apps while keeping their components editable for creators handling complex production work.
Adobe unveiled the preview at its MAX conference in Miami Beach on Nov. 10. The integrations rely on Model Context Protocol (MCP), which Anthropic introduced in 2024 to help AI applications work with external tools and information.
The integrations are designed for agents to make changes to native files on a user’s computer in Photoshop, Illustrator, Premiere and After Effects. Adobe said elements such as layers, vectors, bins and sequences would stay editable after an agent’s work.
Adobe says the tools are intended to give creators more choice and help with multistep work while they retain hands-on editing. The preview adds local project-file access to Adobe’s separate effort to bring creative tools into third-party AI platforms.
Adobe’s own Creative Agent powers beta AI Assistants in Illustrator, Premiere, InDesign, After Effects and Frame.io. The Photoshop AI Assistant is available through the Photoshop beta app. Those assistants are separate from the local connections for third-party agents.
Adobe and Google plan to expand Adobe connectors in Gemini and Gemini Enterprise to include Acrobat and more creative tools. The companies also said they would bring Google’s image and video models to Firefly, Photoshop, Premiere and Adobe Express.
In June, Adobe said it was bringing its creative tools to AI platforms including ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Slack. At MAX, the company also highlighted Relight and Generative Restyle in Photoshop, Color Mode in Premiere and Open Eyes in Lightroom. Adobe lists Open Eyes and InDesign Co-editing among its beta features. The local MCP integrations are planned to enter private beta in early 2027.
