Adobe AI plugins for designers: the short answer

Adobe AI plugins for designers now put Photoshop, Express, Illustrator, Lightroom, Firefly and Acrobat inside the three big assistants. On 24 September 2026 Adobe brought its tools to Gemini and expanded the Adobe for Claude plugin to more than 80 tools, and on 29 September 2026 it added interactive editing controls to the Adobe plugin in ChatGPT: a brush selection, dynamic sliders and a side panel editor that syncs back into the chat.

The practical way to use them is not "ask the chatbot to design something". It is to let the chat handle the repetitive production layer (resizing, retouching batches, adapting one design to many channels, cleaning PDFs) while you keep the decisions about hierarchy, brand and copy. This playbook shows which assistant to open for which job, a repeatable workflow, where it breaks and what it means for the role.

What Adobe actually shipped in each assistant

Here is what Adobe states, without embellishment. Availability and limits vary by account, and Adobe itself notes that responses should be checked.

Adobe in Gemini. Photoshop, Lightroom, Express and Firefly, connected through Personal Intelligence: you select Adobe under connected apps and sign in with an Adobe account (a free one works). Adobe describes three jobs: preparing product photo collections (lighting, color, consistent crops), making social-ready versions of one design while keeping branding, and generating Express marketing templates from a campaign and audience description. It is rolling out globally on all Gemini plans, and Adobe calls it a first step with more tools planned.

Adobe for Claude. More than 80 tools across Acrobat, Express, Photoshop, Illustrator, Premiere, Lightroom, InDesign and Adobe Stock in one plugin, in Claude and Claude Code on desktop, mobile and web. You can use it as a guest, but signing in unlocks higher usage limits, more tools and continuity across sessions. The September update added an Express layer-based editor (change images, copy, colors and fonts on individual elements without regenerating the whole design), a PDF editor, a Document Review skill and new Acrobat tools for organizing, redacting and converting PDFs.

Adobe plugin in ChatGPT. An image editor powered by Photoshop and Adobe Express opens in a side panel. You brush over an object or area, ask for a change, and the editor surfaces controls that match the edit, such as hue, saturation, contrast and brightness sliders scoped to your selection. A PDF editor powered by Acrobat lets you edit text and highlight in place. Edits carry back into the conversation, so you do not re-upload or re-explain. It began rolling out globally on ChatGPT web and app on 29 September 2026.

Which assistant for which design job

Verdict: open ChatGPT when one image needs a precise, visible fix. Open Claude when the work spans many files or formats, or when a design needs element-level edits without regeneration. Open Gemini for product photo collections and quick campaign templates, especially if your team already lives in Google tools.

The Chat, Select, Finish workflow

The pattern that fits all three is a three-step loop: Chat, Select, Finish. It works for social campaigns, product imagery, pitch decks and client deliverables.

Step 1: Chat the brief, not the pixels

Start with constraints the tool cannot guess: the channel, the sizes, the brand colors and fonts, what must not change, and who the audience is. "Adapt this launch visual to Instagram story, LinkedIn banner and a 1200 by 628 ad, keep the logo position and headline untouched" is a usable brief. "Make this pop" is not. If you keep brand rules in writing, paste the relevant part every time; the plugin does not know your guidelines unless you tell it. Teams that already document a design system for AI will have that text ready.

Step 2: Select instead of regenerating

When a result is close, stop prompting the whole thing again. In ChatGPT, brush over the exact area and use the sliders. In Claude, use the layer-based editor to change one text box, one color or one image. This is the real shift in these updates: Adobe moved from "generate and hope" to direct manipulation inside the chat, the same direction covered in ChatGPT Images 2.5 for designers.

Step 3: Finish where the craft lives

Treat the chat output as a strong draft. For anything that ships, open the file in the full app or your design tool, check type, spacing, color profiles and export settings, and name the file properly. For client reviews, Claude's PDF editor and Document Review skill are useful to collect comments, but a human still decides which feedback to accept.

Prompts that work well

• "Use the Adobe plugin to match the lighting and white balance of these six product photos to the first one, then crop all to 4:5 with the product centered."

• "Create three social versions of this design for story, square and banner. Keep the headline wording and logo unchanged. List anything you had to move."

• "In the Express editor, change only the CTA button color to our accent color and leave everything else as is."

• "Review this proposal PDF, highlight every sentence that mentions timeline or payment, and add a comment where wording is unclear."

Asking the assistant to list what it changed is a cheap habit that catches silent edits.

Where it breaks

It is not a design system. The plugins edit assets; they do not hold components, tokens or interaction states. For interface work, tools that write to a real canvas, such as the Figma agent for designers, remain the better fit.

Limits and tiers differ. Guest use in Claude has lower usage limits and fewer tools, and Adobe notes that compatibility and availability vary. Do not plan a deadline around a feature you have not seen on your own account.

Brand drift is silent. A resized variant can shift a font weight or crop a logo margin without saying so. Check every variant against the master before export.

Client files and privacy. Connecting an Adobe account to an assistant means client assets pass through two services. Check your contracts and your clients' AI policies before uploading unreleased work, and use redaction for sensitive PDFs.

Hype vs reality

The hype says Photoshop is now a chat window and anyone can do a designer's job by typing. The reality is narrower and more useful. These updates make the boring middle of design production faster: variants, retouch batches, PDF cleanup, small layout fixes. They do not decide what the campaign should say, which image earns attention or whether the layout supports the message. Verdict: a real productivity gain for production work, close to zero change for the thinking that clients pay a senior designer for.

Where this takes design work

Over the next one to three years, creative tools stop being destinations and become capabilities that appear wherever the conversation is happening. Adobe is now inside ChatGPT, Claude and Gemini at once, which tells you the vendor expects people to start creative work in an assistant and finish in the app. For designers, that means two skills grow in value: writing precise briefs that machines can execute, and reviewing machine output quickly and ruthlessly. The same assistant-first pattern ran through OpenAI DevDay 2026 for designers.

The fear: am I falling behind?

If most of your week is resizing, retouching and exporting variants, that work is getting cheaper, and pretending otherwise does not help. The honest move is to automate it yourself before someone else decides it is all you do, and to spend the recovered hours on concepts, systems and product thinking. The longer answer is in will AI replace UX designers.

Moves for this week: connect Adobe in the assistant you already use, run one real batch job (a set of product photos or a set of social variants) through the Chat, Select, Finish loop, and write down where it drifted. That note becomes your team's checklist.

If you want a product, a launch campaign or AI features designed with this kind of workflow behind them, DIGCY offers UI/UX design services and motion design and AI video. DIGCY is a Dribbble Selected Agency and a UI/UX agency in Casablanca working with clients globally.

FAQ

What can the Adobe plugin do in ChatGPT?

Since 29 September 2026, the Adobe plugin in ChatGPT opens an image editor powered by Photoshop and Adobe Express in a side panel. You brush over an area, describe the change, and refine it with sliders such as brightness and contrast. A PDF editor powered by Acrobat lets you edit text and highlight passages, and edits sync back into the chat.

How do I connect Adobe to Gemini?

In Gemini, open Personal Intelligence, select Adobe under connected apps and sign in with an Adobe account, or create a free one. Adobe says it is rolling out globally on all Gemini plans with Photoshop, Lightroom, Express and Firefly.

Which Adobe apps work in Claude?

The Adobe for Claude plugin includes more than 80 tools across Acrobat, Express, Photoshop, Illustrator, Premiere, Lightroom, InDesign and Adobe Stock. It works in Claude and Claude Code on desktop, mobile and web, as a guest or signed in for higher limits and more tools.

Which assistant is best for designers using Adobe tools?

ChatGPT is best for precise single-image edits thanks to brush selection and scoped sliders. Claude has the widest toolset and a layer-based Express editor, so it suits multi-file and multi-format work. Gemini is focused on product photo collections and quick Express templates.

Will Adobe in ChatGPT replace graphic designers?

It replaces some production tasks, such as resizing, simple retouching and making channel variants. It does not replace deciding what a design must communicate, keeping a brand coherent or judging what will work for an audience, which is where designers add the most value.

Are Adobe AI plugins overhyped?

Partly. They do not turn a chat window into a full design tool. The genuinely useful part is direct editing inside the conversation, which makes repetitive production work faster and keeps the designer in control of the final result.