Gemini ‘Projects’ feature seen in Google app

Google is getting ready to launch a “Projects” workspace in Gemini, suggesting an effort to make its assistant better at managing large, multi-file jobs—much like what OpenAI has done with ChatGPT Projects.

Clues inside the Google app hint at Gemini Projects workspace

An APK teardown was made on the Google app (version 16.37.46.sa.arm64). There’s a new entry in the Gemini side menu under the ‘My Stuff’ section — Projects. The tag isn’t live for general users just yet; all we have to go on are these new label and accompanying text that may hint at the initial user flow involving creating projects through the Gemini web app, with mobile as a browsing and interacting surface.

This staged release is typical of Google’s approach to Gemini features: ship on the web (for control and iteration) first; then bring it to Android once the core model behaviors and UI are locked down. Anticipate feature flags, A/B testing, and a slow roll-out rather than a big-bang launch.

What the new Gemini ‘Projects’ workspace likely enables

Projects are like smart workspaces—they automate all the machinery around keeping files, links, and instructions for your project while preserving context throughout your days, weeks, months, or even years working on a project. ChatGPT Projects from OpenAI Docs has something very much like this: scoped workplaces with specific instructions and documents attached which the model can refer to without having to be re-prompted every time.

If Gemini follows that template, we might see users generate a project for a product launch, append a brief, assets, and constraints to it, and click send with this assistant and persistent rules (“maintain hold to B2B tone; reference pricing policy; use brand glossary”). Every interaction would receive that context, increasing consistency and minimizing the need for prompt engineering.

For researchers, a project may package interview transcripts, PDFs, and web sources that allow Gemini to summarize, compare, and draft memos with citations. Teachers might collect lesson plans, standards, and rubrics, and then generate differentiated materials without having to re-create context for every class.

NotebookLM-style capabilities embedded directly in Gemini

Google currently operates a similar-sounding entity in NotebookLM, which taps into data provided by users to create a knowledge-grounded notebook. The difference here is in placement: Projects will live directly inside Gemini’s main experience instead of being a standalone app, lowering the friction and potentially expanding usage.

Adding Projects in Gemini also makes a natural bridge to Google Workspace. Once Projects can simply reference files in Drive, Docs, and Sheets, it becomes quite a useful hub for teams. Even without document automation, pinning a curated context to a conversation is going to be such an obviously better output that it’s something enterprise buyers constantly request in analyst briefings when dealing with reticent firms like Gartner.

Competitive pressure from ChatGPT is shaping Google’s plans

The ushering in of Projects by OpenAI reconfigured how an assistant thought about ongoing work — not so much one-off chat, but durable spaces ready for teammates. By adding a like tool to Gemini, Google is able to fill a usability hole for power users, such as those managing several initiatives or who are working cross-functionally.

The timing is notable. Gemini has added grounding over time with the help of extensions and integrations, but long-lasting themed contextualization is what takes an assistant from “helpful” to “trustworthy.” It helps reduce hallucinations by confining the reference set and promotes workflows that can be reproducibly followed by an organization.

What was spotted first about Gemini Projects in the app

Projects in the scope of Gemini were first spotted by TestingCatalog on X, which came across early mentions of the feature in app strings. That the experience is now showing up inside the Google app’s UI is an encouraging sign that a user-facing launch may be on the horizon, despite web-only production initially.

What to watch next as Gemini Projects approaches launch

Three key lingering questions:

• Will Projects support team sharing, roles, and permissions?
• Can it draw on structured data in Sheets or entries, summaries from Docs and PDFs in Drive?
• Is there something like project-level custom instructions (overrides?) compared to global Gemini settings?

Those are the details that will decide whether Projects is a personal org or a believable collaboration layer.

If Google marries Projects with strong source grounding and integrates it deeply into Workspace, Gemini is no mere chat box—it’s a work surface for active work.

That’s the standard modern AI assistants have set, and that’s precisely where user demand is moving.