On November 10, 2026, at the Adobe MAX conference, Google and Adobe announced an expansion of their partnership, connecting Adobe's creative stack with Google's office and search stack. Acrobat will receive an 'Edit with' button in file previews in Google Drive and Gmail, Adobe connectors will appear in Gemini and Gemini Enterprise, and Google's front-end models—Gemini Omni, Veo, and Nano Banana—will be integrated into Firefly, Photoshop, Premiere, and Adobe Express. Separately, Adobe and Google DeepMind will begin joint work on an image model for Photoshop capable of making edits in near real-time. The deal is based on Adobe's scale: over 1 billion monthly active users and more than 400 billion PDFs opened annually.

image
image

What Happened

The announcement was made at the Adobe MAX conference on November 10, 2026. Acrobat will become a recommended application in Google Drive and Gmail: an 'Edit with' entry point will appear in file previews, opening the document in Acrobat without leaving the service. Adobe connectors in Gemini and Gemini Enterprise will connect Acrobat and additional creative tools; these tools are also promised in AI Mode in Google Search. In the reverse direction, Google's front-end models—Gemini Omni, Veo, and Nano Banana—are being embedded into Firefly, Photoshop, Premiere, and Adobe Express, and Gemini models will power the AI Assistant in Acrobat for conversational PDF editing. Another direction is the joint development by Adobe and Google DeepMind: an image model for Photoshop designed for near real-time editing. This is currently an announcement of intent: the listed features are not yet available in partner applications at the time of publication.

Context

In terms of content, this is a distribution swap rather than a technological breakthrough: the models in question already exist, and the announcement contains no new architecture, training method, or benchmarks. For Adobe, the deal confirms a strategy of relying on multiple model providers; for Google, it opens the way for front-end models to enter professional creative applications, bypassing its own consumer services. The working pattern can be described as 'models in the app, app in the platform': the entry point in the Drive preview and connectors within the Gemini chat turn Adobe applications into a set of callable actions within Google ecosystems. Adobe's scale figures are read in an ML context primarily as an argument for telemetry: the volume of real user editing sessions is a valuable signal for model training and iteration. In terms of detail level, the announcement remains a demonstration of intent rather than an engineering specification.

Why This Matters for the Industry

The announcement merges two major stacks—creative and office/search—and outlines a battle for default actions within platforms: what matters is not the tool itself, but which action the platform offers by default in the Drive preview, in the Gemini chat, or in the AI Mode response. For startups, this is a signal of space compression: scenarios for conversational PDF editing and prompt-based image editing are being closed off by flagship products, and thin overlays around these tasks lose their meaning. The research value is minimal; the only technically substantive claim is the joint image model for Photoshop from Adobe and Google DeepMind. There is more potential in the AI Assistant in Acrobat as a document AI testing ground: if the integration is released, it will be possible to independently verify the accuracy of local edits without breaking the layout, behavior on scans, and the actual latency of conversational editing on a stream of ordinary user documents.

Why This Matters for Users

Readers are promised two changes in familiar places. First—documents: a significant part of routine PDF work will no longer require downloading the file and switching between tabs, because the preview in Drive or Gmail will turn into sending the document for editing. Second—tools: Adobe connectors within Gemini and in AI Mode in search mean that some of Adobe's creative functions can be invoked near where the user is already searching for information or talking to an assistant. Professionals in Photoshop are promised prompt-based edits from the joint image model with DeepMind. The practical benefit is currently zero: the announcement was made today, and everything will change only with the actual rollouts of individual features.

What Is Still Unknown / Limitations

Sources do not specify dates for the rollout of individual items, APIs and pricing, or measurements of conversational PDF editing latency. The claimed 'near real-time' mode in Photoshop is not supported by verifiable estimates: there are no metrics for edit quality, the name of the base model and its version, and the joint work between Adobe and DeepMind is only beginning. It is unclear which integrations will reach users and in what order. The cited coverage figures are data from the companies themselves; there is no independent verification in the materials. There are also no technical reports, benchmarks, or public demonstrations of the claimed scenarios in the sources.

Sources

Author

Look at AI, editorial team