Technology

Adobe Adds More AI to Its Creative Apps, and Lets Outside Chatbots Plug In

Martin HollowayPublished 5d ago4 min readBased on 1 source
Reading level
Adobe Adds More AI to Its Creative Apps, and Lets Outside Chatbots Plug In
source:adobe.com

Adobe has announced new AI features for Photoshop, Premiere, Illustrator, and InDesign. It has also released a tool that lets outside chatbots use Adobe's software. The announcements were made at Adobe's MAX event in Miami, according to TechCrunch.

The chatbot connection is the most unusual part of the release. Adobe calls it a Model Context Protocol (MCP) server. Think of it as a standard power socket. Any chatbot built to use that socket can connect to Adobe's tools. ChatGPT, Claude, and Gemini are named as examples.

Adobe has two versions. The online version is for general-purpose assistants. The local version runs on the user's own computer, where the desktop applications are installed, so its requests stay on that machine. It covers Illustrator, Photoshop, After Effects, and Premiere.

Adobe's own AI system is called Firefly. It now includes a creative AI agent, which is a program that carries out tasks on its own. Adobe says the agent is proactive and always on, so it can learn more about what a user is working on.

Firefly also has several separate tools:

  • An image editing feature changes one part of a picture and leaves the rest alone.
  • A video tool builds a sequence from clips, pictures, and sound effects.
  • A tool creates charts that match a brand's look.
  • Firefly Builder, a tool for making quick design mockups, is in private beta. That means only a limited group of testers can use it.

The desktop apps also received new features. Photoshop can relight a scene. Its sliders can now adjust the style of AI-generated images. Premiere has a new color grading system for professionals, plus a way to remove objects from video. Lightroom can fix closed eyes in photos and remove blemishes. Quick actions can remove a background or create a video thumbnail in one step.

Adobe also lets users choose AI models from other companies inside its products. These include Google, Runway, ElevenLabs, Black Forest Labs, OpenAI, Kling, and ByteDance. So users are not limited to Adobe's Firefly models.

For professionals, the key questions are about how well the tools fit into their work and how much control they have. Letting an outside program work directly on layers, timelines, and asset libraries is different from copying a picture from one app into another. The announcement, as reported, does not say whether those permissions are tight enough for busy production teams. It also does not explain how the always-on agent handles access to a user's files. IT and creative teams will need to test both points.

The broader context here is a change in what creative software is. Until recently, AI features in these programs were mostly small panels inside the app. Now the programs open their functions to any compatible chatbot. The software is becoming a set of doors that other programs can walk through. That is a change in how the product is built, not only in what it can do.

In my view, the real test of this release will come in everyday work, not in the keynote demo. Editors and photo retouchers have learned over many years where automatic fixes break down. A relighting slider earns trust only after it handles a hard photo well. The always-on agent also raises an old question about how much a tool should observe before anyone asks for help. The answer will depend on the controls Adobe offers, and those details will matter as the features leave private beta.

Even so, the possibilities are easy to picture. A designer could ask a general chatbot to make a thumbnail, retouch a set of portraits, or put together a short video from a folder of clips. The work would happen inside the professional program where the files already sit. Whether this saves real time will depend on the people who use it every day.