Loading page…
A local MCP endpoint gives external assistants access to creative-app capabilities. Adobe says user feedback will help determine how much it exposes.
Listen to this story
Adobe is extending its creative software beyond its own interfaces: a local Model Context Protocol (MCP) endpoint will let assistants including ChatGPT, Claude and Gemini call capabilities in Photoshop, Illustrator, After Effects and Premiere. The company says the initial endpoints are a starting point, and vice president Deepa Subramaniam says user feedback will shape what becomes available, so the breadth of access is not yet settled. At MAX in Miami, Adobe also detailed Firefly creation tools and hands-on controls in its editors, signaling a mix of agent access and targeted, adjustable AI features rather than a chat-only workflow.
Firefly’s Builder, currently in private beta, is a prototyping tool for turning ideas into demonstrable designs.
Firefly will add video assembly from clips, images and sound effects, plus branded graph creation and localized image editing.
Adobe describes its always-on Firefly agent as a way to give it more context about ongoing work; the announcement gives no release timing.
Adobe is adding AI controls inside its creative apps while giving outside chatbots a wider route into those tools. At MAX in Miami, the company announced coordinated Creative Cloud and Firefly updates, including a local connection that exposes capabilities across Photoshop, Illustrator, After Effects and Premiere to external assistants.
The connection uses Model Context Protocol, or MCP—a way for chatbots to access functions in other software. Adobe is releasing a local MCP endpoint covering capabilities in four creative apps. It provides another route to Adobe functions for assistants such as ChatGPT, Claude and Gemini, alongside the new features inside Adobe’s own products.
Adobe already has ChatGPT and Claude integrations that expose some product functions. The new endpoint is intended to expose more Creative Cloud capabilities, rather than introduce chatbot access for the first time.
Its scope is still taking shape. Adobe vice president Deepa Subramaniam told TechCrunch that the company will use feedback to decide which capabilities to expose. She described the initial endpoints as a starting point for learning what people want to do, not a settled boundary around the apps’ functions.
Inside Firefly, Adobe’s announcements cover several stages of creative production. Builder, a prototyping tool in private beta, lets users create designs to demonstrate ideas.
A new video tool will assemble clips with other elements, including images and sound effects. Adobe also announced a tool for on-brand graphs and an image editor designed to change one part of a picture without changing other objects.
Adobe is also making Firefly’s agent proactive and always on, with the stated aim of giving it more context about ongoing work. Alongside that broader assistant role, the company is simplifying smaller actions, including removing a background and generating a video thumbnail.
Adobe is also adding controls directly to its established editors. Rather than presenting every task as a conversation with an assistant, these updates include sliders, a color-grading system and tools for specific corrections in photos and video.
Loading discussion...
Join the conversation
Explain when proactive help would feel useful rather than intrusive.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.