On November 10, 2026, at its MAX event in Miami, Adobe unveiled a broad round of AI upgrades across Photoshop, Premiere, Illustrator, InDesign and Firefly. The headline is the Adobe Firefly AI agent, built to stay on in the background, along with an MCP server that lets outside chatbots call Creative Cloud features directly. For businesses and software teams, it signals that creative tools are moving from "an AI button added on" to "software an agent can drive".
The Firefly agent no longer waits for a prompt
According to TechCrunch, Adobe is turning its creative assistant into a proactive, always-on agent. The goal is for it to pick up context about what you are working on and be ready to help at any step, instead of answering only when summoned.
Quick actions are getting simpler too. Removing a background or generating a video thumbnail now takes a single step. It fits the agent trend we see across many platforms, such as Claude Personal Finance analyzing spending through Plaid: AI moving into the real workflow rather than sitting in a separate chat window.
Builder: Firefly steps onto Figma's turf
Inside Firefly, Adobe is testing a tool called Builder, currently in private beta. Users can put together designs to show off an idea, which is essentially prototyping, long Figma's home ground.
Firefly also gains:
- A tool for creating on-brand charts and graphs.
- An image editor that changes one part of a picture without touching other objects.
- An AI video tool that assembles clips, images and sound effects into a finished video.
Taken one by one, none of these is a breakthrough. Together, they show Adobe wants Firefly to be where a project begins, not just a place to generate images.
What Photoshop, Premiere and Lightroom get
The familiar apps receive their own additions:
- Photoshop: relighting, plus sliders for prompt-based generative styles.
- Premiere: a new color grading system for professionals and object removal for video.
- Lightroom: a fix for blinking eyes and a blemish removal tool.
Adobe is also doubling down on a multi-model approach. It already offers models from Google, Runway, ElevenLabs, Black Forest Labs, OpenAI, Kling and ByteDance. Topaz Labs, a recent Adobe acquisition, is releasing two image models: Wonder 4 for upscaling and denoising, and ImageHDR for turning standard images into HDR.
A local MCP so ChatGPT, Claude and Gemini can use Creative Cloud
The most technically meaningful piece is the local Model Context Protocol (MCP). Adobe already had integrations with ChatGPT and Claude that expose some functions. The new MCP opens up more of Creative Cloud to chatbots like ChatGPT, Claude and Gemini, with endpoints across Illustrator, Photoshop, After Effects and Premiere.
Adobe vice president Deepa Subramaniam told TechCrunch the company is still deciding which capabilities to expose and will lean on user feedback. In other words, the scope of the MCP will keep shifting with how customers use it.
A dissenting view is worth noting: last month, Figma co-founder Dylan Field argued that MCP is not a perfect standard and benefits frontier labs more. The protocol remains a subject of debate.
What it means for businesses and software teams
If you build digital products, three things deserve attention.
Design workflows can be automated by agents. Once a chatbot can call Photoshop or Premiere through MCP, repetitive jobs such as normalizing product photos, cutting variants for different channels or assembling short videos can be handed off. You still need human review, especially for brand content.
MCP is becoming the default doorway into software. If a heavyweight like Creative Cloud exposes endpoints for agents, internal systems such as CRM, ERP and project tools will soon face the same expectation. We covered the lightweight-agent trend in Meta's Muse Ember: a small AI model for agent tasks, and MCP is the connective layer that lets such agents work with real software.
The line between design and coding keeps blurring. Firefly's prototyping tool meets the wave of vibe-coding apps, such as Automattic's Spacefast. An idea can now travel from sketch to working prototype in very little time.
What to prepare right now
You do not need to overhaul your process, but a few small steps make sense:
- List the repetitive creative tasks that eat the most hours on your team.
- Pilot MCP on one small workflow with a human reviewing the output.
- Decide which data may go into external chatbots, particularly brand assets and client documents.
- For internal systems, consider opening MCP endpoints with proper access control.
Keep in mind that Builder is still in private beta and Adobe has not fixed the MCP's scope, so treat this as a direction to test, not a platform to bet heavily on yet.
Conclusion
Adobe is betting that the future of creative software is a proactive, multi-model agent that any chatbot can call. For businesses, the question is no longer whether to use AI, but whether your own systems are ready for agents to plug in.
If you are considering AI agent integration or exposing an API/MCP layer for internal software, contact the Besoft team for tailored advice.
Sources
- TechCrunch, "Adobe adds more AI to its products and launches an MCP", Ivan Mehta, November 10, 2026: https://techcrunch.com/2026/11/10/adobe-adds-more-ai-to-its-products-and-launches-an-mcp/
