Adobe on September 24, 2026 began a global rollout of tools from Photoshop, Lightroom, Adobe Express and Firefly inside Google's Gemini assistant, and in the same announcement brought Acrobat into its plugin for Anthropic's Claude, which the company says now carries more than 80 tools. The Gemini release fills a gap left by Google a day earlier, when it named Adobe as a partner without saying which Adobe products would be connected.

In Short

Adobe has made some of its photo and design tools usable inside Google's Gemini chatbot, and has added its PDF tools to the Adobe add-on that already worked inside Anthropic's Claude. That matters to marketers, designers and anyone who handles documents, because retouching product photos, resizing an ad for Instagram or blacking out names in a contract can now start with a typed request in a chat window rather than inside Adobe's own apps. What changes is where that work begins, although Adobe has not listed excluded countries, explained what happens to uploaded files, or said where its figure of more than one billion monthly Gemini users comes from.

What Adobe connected to Gemini

The details sit in a post on Adobe's blog by Deepti Pradeep, dated September 24, and on a product page for what Adobe calls Adobe for Google Gemini. The blog itself uses the shorter name Adobe in Gemini. Through it, according to Adobe, a Gemini user can call on tools from four applications - Photoshop, Lightroom, Adobe Express and Firefly - to enhance photos, create marketing assets and customise designs. The user describes a result. Adobe picks the tools and the order in which they run.

0:00
/0:50

That settles a question left open on September 23, when Google began rolling out 14 new Connected Apps partners and listed Adobe only as a company, with no indication of whether the connection ran through Express, Photoshop, Acrobat or something else. Acrobat, it turns out, is not part of the Gemini release.

Adobe's blog names three uses. The first is product photography: a user uploads shots and describes the look, and Adobe adjusts lighting and colour, crops consistently and prepares the set for a storefront, website or campaign. The second turns one design into social-ready versions that keep branding and messaging intact. The third starts from Express templates. A user describes a campaign, a target audience and a style; Adobe surfaces matching designs; text, imagery and colours are then personalised on request.

The product page goes further, describing a Lightroom preset applied across a batch of product photos, a logo file dragged into the conversation to brand mockups of several products at once, and image variations for ecommerce, Instagram, TikTok and Facebook that can then be downloaded. Express templates can be customised and animated for social media. A promotional image at the top of Adobe's post shows the invocation in practice: a Gemini prompt bar reading "@Adobe give my pics a consistent look and feel", the same @-mention convention Google described for its connected partners.

Google supplied a statement for the post. "We're excited to expand our partnership with Adobe and bring industry-leading creative tools directly into Gemini. Now, anyone can create, edit and bring their ideas to life right in the chat," said Neal Pancholi, Managing Director, Global Partnerships, GDM & Gemini, at Google.

Adobe presents the release as an opening set rather than a finished one. "As we've done across other AI platforms, we'll continue to enhance the experience with more tools for creativity and productivity," the post states. No dates or product names accompany that commitment, and Adobe gives no tool count for Gemini.

A billion users, or 950 million?

Adobe describes Gemini as helping "more than one billion people each month". That runs ahead of the latest figure Alphabet has published. The company reported 750 million monthly active users for the Gemini app on February 4, a number that crossed 900 million in May and reached 950 million in the second quarter. Neither Adobe's post nor Google's quoted statement gives a date, a source or a definition for the higher figure, which may reflect growth since July or a different measure altogether.

Setup, accounts and a discrepancy

According to Adobe, the Gemini integration is available on every Gemini subscription tier and is rolling out globally from September 24. The blog sets out the route: open Personal Intelligence, select Adobe under connected apps, then sign in with an Adobe account or create a free one.

Adobe's product page describes five steps - open Gemini, connect Adobe, sign in, prompt in chat, start creating - yet also states that no Adobe account is needed to get started, with a login unlocking more capabilities, storage for assets and continuity across sessions. The two documents, published by the same company on the same day, therefore differ on whether signing in is part of setup or an optional upgrade.

Google's September 23 post described a different path, through Gemini settings or an @ mention, and its help pages add a prerequisite that Adobe's instructions omit: Connected Apps on gemini.google.com, on iOS and on smart watches stop working when Keep Activity, the setting formerly called Gemini Apps Activity, is switched off.

The footnote to Adobe's post reads: "Check responses. Setup required. Compatibility and availability varies." The word globally and the phrase availability varies sit a few paragraphs apart in the same post, and no excluded country is named. European readers have reason to check. Documentation for Gemini Spark lists the European Economic Area, Nigeria, Switzerland and the United Kingdom as unavailable markets, and Adobe does not say whether its integration follows that map, the Connected Apps map, or neither.

Account type is the other unknown. Google's August availability table for Connected Apps showed most outside services requiring a personal Google Account, with apps on work and school accounts enabled or disabled only by a Workspace administrator. Adobe's reference to every Gemini plan does not say whether Workspace business editions are included. For agencies that run client work on Workspace identities, that is the operative question, and neither company has answered it.

Acrobat joins the Claude plugin

Claude has hosted Adobe's creative tools since earlier in 2026, through what Adobe calls its Adobe for creativityconnector, covering image, vector, layout and video editing. The September 24 update adds Acrobat. "We're adding Acrobat tools to our plugin for the first time, bringing them together with our existing creative tools in one unified Adobe plugin," Adobe's post states.

The combined Adobe for Claude plugin now reaches more than 80 tools across Acrobat, Express, Photoshop, Illustrator, Premiere, Lightroom, InDesign, Adobe Stock "and more", according to Adobe. The product page lists nine applications, Firefly among them. Gemini, by comparison, gets four.

0:00
/2:50

The Acrobat functions are document housekeeping: deleting, combining or splitting pages, redacting sensitive information, and converting or exporting PDFs to other file formats. The product page adds turning designs into PDFs, editing content and adding annotations.

Two hands-on editors

Adobe pairs the tool count with two interactive editors that open inside Claude. In the PDF editor, pages can be reordered, text edited, and documents marked up with comments and highlights. Alongside it sits a new Document Review skill which, according to Adobe, asks Claude to analyse a document and place recommendations and notations directly in the editor.

The second is a layer-based editor for Express designs. Images, copy, colours and fonts can be selected and changed independently, according to Adobe, without regenerating the entire design. In production terms the distinction is not trivial. A text prompt that alters one element of a generated layout can disturb the rest of it; a layer model holds everything else in place.

Both editors sit on top of functions carried over from the earlier connector: refining images singly or in batches, building vectors and layouts, editing video and adapting content for different channels. Adobe's product page offers sample prompts, among them "Balance the lighting and auto-straighten all these photos" and "Resize my video for YouTube, TikTok, and my Instagram feed." It also describes licensing and editing Adobe Stock images and video without leaving the conversation.

Evidence for the benefit is anecdotal. According to Adobe's post, feedback since the original Claude release has included a photographer refining light, colour and consistency across a photo series, and a creative director who significantly cut the time required to develop an eco-fashion campaign. Neither person is named. No time saving is quantified.

Plugin, connector, skill

The packaging says something about how these integrations are now built. Adobe's Claude product page lists four steps: open Claude, install the Adobe plugin, install the Adobe connector, launch the Adobe skill. A connector, in Claude's vocabulary, is the product wrapper around a server that speaks the Model Context Protocol, the open standard through which an AI application discovers and calls outside tools. A plugin is the larger package, a bundle rather than a single-purpose extension in 2026 usage, and Adobe's four-step list follows that shape: one package holding a connection and a skill.

According to Adobe, installations of the earlier connector update automatically to the new plugin, and the rollout is global, covering Claude and Claude Code, Anthropic's coding agent, on desktop, mobile and web. The plugin works for guests. Signing in with an Adobe account unlocks higher usage limits, additional tools and continuity across sessions, according to the blog; the product page adds more skills, generative AI tools and storage for assets. Neither document quantifies a usage limit, and neither names the Claude subscription tiers on which the plugin runs.

What the documents leave open

Data handling is the largest gap. Adobe's materials do not say how long files uploaded through either assistant are kept, whether they train models, or how they are deleted. The only relevant detail is that signing in provides storage for assets. Both product pages carry FAQ entries including "Can I use my own files and Creative Cloud assets?" and "Do I need an Adobe account or paid subscription?", but the versions reviewed show the questions with the answers collapsed.

On the Gemini side, Google's privacy documentation already sets one boundary. Disconnecting an app does not remove data from Gemini activity, and partners keep what Gemini shares with them under their own policies. Adobe's post adds nothing to that picture for its own integration.

The Document Review skill raises a separate, technical issue. A PDF handed to Claude for analysis is content the model reads as data. Prompt injection is the class of attack in which such content is executed as an instruction instead, and contracts, briefs and proposals routinely arrive from outside parties. Adobe's materials describe what the skill does, not how it treats instructions embedded in the documents it reviews.

Commercial terms are absent too: neither company says whether money changes hands for the Gemini listing, and Adobe does not say whether generations inside either assistant count against any Adobe allowance.

Five platforms, one distribution plan

Gemini is the fifth AI platform Adobe names. "We've already brought Adobe to ChatGPT, Claude, Slack and Copilot," the post states. The sequence behind that sentence runs back more than two years. Adobe added an AI Assistant to Reader and Acrobat in February 2024, put a conversational assistant into Adobe Express on October 28, 2025 on a platform built to integrate with chatbots, and turned Acrobat into a productivity agent on May 6, 2026.

The Claude update mirrors an arrangement Adobe already had in ChatGPT. When Semrush, now an Adobe company, published its Claude connector on August 26, the announcement pointed to a recent Adobe plugin for ChatGPT that combined creativity and productivity tools in a single package across ChatGPT Work and Codex. OpenAI's plugin model itself dates from July 9, when the company folded Atlas into ChatGPT Work and moved to a plugin directory. Claude now has the same combination. Semrush itself came into the group through a $1.9 billion acquisition agreed on November 19, 2025.

Adobe also reaches Claude by a second, enterprise route. At Cannes Lions on June 22, Adobe said its CX skills and MCP servers were generally available inside Claude Enterprise and Microsoft 365 Copilot, alongside deployments of its agentic AI products with Accenture Song, Omnicom, Stagwell's Code and Theory and WPP. That is customer-experience software for marketing departments; the plugin expanded on September 24 is creative and document tooling for individuals. Two Adobe product lines now reach the same assistant through different doors.

Why push so hard into other companies' interfaces? Adobe's own research hints at one reason. A survey of 1,002 US workers shared in July found interest in creative AI high but daily use concentrated in routine text tasks rather than image, video and ideation work. Chat assistants are one of the places where that text work happens. Adobe's post describes the chatbot as a place to begin, with its flagship applications reserved for work needing more depth and precision - an upgrade path that starts, in Claude at least, with a guest session and runs through an Adobe login.

Why this matters for marketers

The first consequence is competitive. In Gemini, Adobe enters a creativity group alongside Picsart, Squarespace, Webflow and Canva, which has been connected since Spark's debut at Google I/O on May 19 and became the design partner when Connected Apps reached AI Mode in July. Canva arrives with scale of its own: 10.5 billion web visits in the twelve months to April 2026, second only to ChatGPT among the AI tools one study ranked. Which app receives a design request is therefore the commercial question. When Connected Apps reached AI Mode, Google's help documentation said the system may default to the last application a user relied on for a given task. Google has published no equivalent routing rules for the Gemini app.

None of that position is bought. Google has said it has no plans for ads in the Gemini app, and Anthropic ran a Super Bowl campaign in February attacking OpenAI's move into ChatGPT advertising. ChatGPT is different, and Adobe sits on both sides of it: among the first advertisers when the ChatGPT ad pilot opened on February 9, and on OpenAI's list of confirmed ad technology partners by May.

The second consequence concerns production. Resizing a campaign video for YouTube, TikTok and Instagram, batch-correcting product shots or redacting a client contract are routine agency tasks, and moving them into a chat window changes who performs them and where the output is checked. Brand governance in that setting rests on the assistant's interpretation of a prompt and on Adobe's claim that social versions preserve branding and messaging. Adobe's post does not mention brand kits, locked templates or approval steps.

The third is measurement. Google Analytics added an AI Assistant channel in May to classify referral traffic from ChatGPT, Gemini and Claude. A video resized inside Claude, or a mockup built inside Gemini, generates no referral at all. For Adobe, the same shift moves activity outside Creative Cloud's own interfaces, and its materials position an Adobe sign-in, with higher limits, storage and continuity, as the step that links the two.

Audience size explains the bet: Similarweb's June update put Gemini at 27.3 percent of generative AI website traffic and Claude at 8.9 percent, up from 1.6 percent a year earlier, and Adobe now sits in both as well as in ChatGPT.

What remains unsettled is documentation. Adobe has fixed the product lists, the tool count and the start of a rollout. It has not fixed which countries are excluded, what happens to uploaded files, whether Workspace accounts qualify, or how Gemini reached a billion users. Which of those gaps will close first, and in whose documentation - Adobe's or Google's?

Timeline

Summary

Who: Adobe, through a blog post by Deepti Pradeep and product pages for Adobe for Google Gemini and Adobe for Claude, with a supporting statement from Neal Pancholi, Managing Director, Global Partnerships, GDM & Gemini, at Google. The host assistants are Google's Gemini and Anthropic's Claude.

What: A Gemini integration giving access to tools from Photoshop, Lightroom, Adobe Express and Firefly for photo enhancement, marketing assets, mockups, resizing and template customisation; and an expanded Adobe for Claude plugin that adds Acrobat for the first time, bringing the total to more than 80 tools, with a new interactive PDF editor, a Document Review skill and a layer-based editor for Express designs. Adobe describes Gemini as reaching more than one billion people a month, above the 950 million monthly active users Alphabet last reported.

When: September 24, 2026, one day after Google named Adobe among 14 new Connected Apps partners for Gemini without identifying which products were involved.

Where: Globally, according to Adobe, across every Gemini plan and across Claude and Claude Code on desktop, mobile and web, though Adobe's own footnote states that compatibility and availability vary and no excluded country is listed.

Why: Adobe is placing its tools inside the assistants where work increasingly starts, following earlier integrations with ChatGPT, Slack and Copilot. For marketers, the change moves routine production and document work into chat interfaces where app placement is allocated rather than auctioned, where referral analytics see nothing, and where data handling, account eligibility and market coverage have yet to be documented.