Gemini in Google Workspace: Cross-App Decks, Access, and Alternatives (2026)
On September 9, 2026, Google announced a change to Gemini in Google Workspace. It can now build a deck, a doc, or a spreadsheet from inside whichever app you are already in. You can ask from Gmail, Drive, Docs, Slides, or Chat. The file lands in your Drive for review.
That is a small sentence with a large consequence for anyone who assembles the same status deck every month. The work used to start with a tab switch. Now it starts with a prompt in the window you already had open.
What Google announced, and when
Google published the change on two official channels on the same day. The Workspace Updates post carries a publication date of September 9, 2026. The companion post on the Google Workspace Blog carries the same date in its page metadata.
The rollout is gradual. Google's update states that Rapid Release and Scheduled Release domains both began a gradual rollout on September 2, 2026. Visibility can take up to 15 days.
The mechanism is described plainly. Gemini "works as an intelligent orchestrator," Google writes, using Workspace Intelligence and deeper app integrations. It gathers context "from your relevant files, emails, and chat threads, depending on which sources you select (or your admin has enabled)."
Five capability groups are named:
- Create content — generate "formatted Google Docs, structured Sheets, or stylized Slides in the background, saved securely to your Drive"
- Conduct deep research — synthesize material across folders and long threads into a summary report with source attributions
- Draft and send emails — compose from meeting notes or the active document, then open a draft in Gmail
- Schedule your meetings — find open times or resolve conflicts without switching to Calendar
- Keep track of your to-dos — log reminders into Google Tasks
What this changes for the deck you have to build
The deck case is the one Google leads with. Its own example prompt, run from Chat, is: "Create a deck outlining Project Zebra with the latest updates." A second example, run from Drive, is: "Create a customer insights deck from project Zebra using my project files including sheets, reports, and emails."
Read those two prompts closely and the design becomes clear. The raw material is whatever already lives in Workspace. Gemini gathers it, drafts the slides, and hands back a link to the presentation.
A third example runs the other way. From Docs, the suggested prompt is: "Turn this proposal into an easy to read slide deck for my director using @presentation as a style reference." The style reference is the interesting part. It points at an existing file rather than at a theme picker.
Google flags one behaviour worth knowing before you try it. Its note on the Chat example says you "might notice Gemini asking some clarifying questions in this experience." The blog also says the presentation flow is "coming soon to all of the places where you can create presentations in Workspace." Coverage is still widening.
Two guardrails are documented. For anything outbound, such as sending mail or booking time, Gemini "presents an interactive preview card, allowing users to review, edit, and confirm before execution." On permissions, Google writes: "If the user cannot access a document, neither can Gemini."
Who can use it today
Google lists availability by edition rather than by price on this page. Here is what the announcement states, as of September 10, 2026.
| Edition group | Access described in the announcement |
|---|---|
| Business | Standard and Plus |
| Enterprise | Standard and Plus |
| Consumer | Google AI Pro (scheduling excluded) and Google AI Ultra |
| Other editions | Frontline Plus, scheduling functionality only |
| Education add-ons | Google AI Pro for Education |
| Other add-ons | AI Expanded Access |
Two limits sit in the same section. Usage of the advanced AI features across Workspace apps is "subject to usage limits." And the launch is English-only: "At launch, this feature will be supported in English only. Support for more languages will be added in the future."
How to turn it on, and what it costs you in limits
For most people there is nothing to install. The capability appears inside Gmail, Drive, Docs, Slides, and Chat as the gradual rollout reaches your domain. Google dates that rollout from September 2, 2026, with up to 15 days for visibility.
What you do control is the source list. The announcement is explicit that Gemini gathers context from files, emails, and chat threads "depending on which sources you select (or your admin has enabled)." That sentence is worth reading as a setting rather than as marketing. If the deck comes back thin, the first thing to check is which sources were in scope, not which prompt you used.
Three limits are documented alongside the launch, and all three matter more than they look.
Usage is metered. Google states that use of the advanced AI features across Workspace apps is "subject to usage limits." The page does not publish a number, so treat heavy daily use as an open question rather than a settled one.
English only at launch. The announcement says support for more languages will be added in the future. For a team that reports in two languages, that decides where this fits today.
Scheduling is carved out on some plans. Google AI Pro is listed with scheduling excluded, and Frontline Plus is listed as scheduling only. If calendar actions are the reason you want this, check the edition row before you promise it to anyone.
What about privacy
Two statements in the announcement cover the common questions. On training, Google writes that your data "is not reviewed by humans or used to train Gemini models." On access, the rule is inheritance rather than a new permission layer: "If the user cannot access a document, neither can Gemini."
That second line is the one to repeat internally. It means the blast radius of a bad prompt is bounded by the permissions the person already had. That is a materially different risk profile from a tool that indexes everything centrally.
Where a cross-app assistant stops helping
Everything above depends on one condition: the material has to be inside Workspace already, or in a source your admin has connected. That is a reasonable design, and for teams that live in Drive it removes real friction.
It is a different starting point from the one many analysts actually have. A quarterly export lands in your downloads folder. A vendor emails a workbook with four tabs and a merged header row. A finance system produces a CSV that nobody has cleaned yet. In those cases the question is not which app you are in. It is what the numbers in that file say, and whether you can show your work when someone challenges a figure.
That is the gap Powerdrill Bloom is built for. You upload the file and ask in natural language. Its homepage describes the product as "the AI data analyst that shows its work." It also states that "every number comes back with the page, the row and the figure behind it." From there the same workspace can produce slides, docs, sheets, and images.
This is also where the deliverable pages on our own site fit. If the job ends in slides built from a workbook, Excel to PPT covers that path. The topic to PPT page covers the case where you start from a subject instead.
The two approaches are complements more often than substitutes. Use the Workspace assistant when the context is already in Workspace. Reach for an upload-first agent when the context arrives as a file.
Why the deck example is the one to watch
Of the five capability groups, four automate something you were going to do anyway: an email, a meeting, a reminder, a summary. The deck is different, because building one is where most of the week actually goes.
Google's Drive example is the honest version of the pitch. It reads: "Create a customer insights deck from project Zebra using my project files including sheets, reports, and emails." Three source types, one artifact, no tab switching. If that works reliably on your material, it removes an afternoon rather than a minute.
The caveat is in the same sentence. It works on your project files, meaning the ones already sitting in Drive with the right sharing. Gemini in Google Workspace is bounded by that condition by design. It is also the condition most analysts break the moment somebody emails them a workbook.
Alternatives worth knowing
- Claude Cowork — an agent you hand a goal to. Anthropic's product page says it "works across your files and tools" and "works directly in folders and tools you choose," then returns work for review.
- Bricks — file-first by design. Its own FAQ describes it as "an AI data analyst that instantly creates beautiful dashboards and reports from your CSV, Excel, or PDF data."
- Powerdrill Bloom — an upload-first workspace that answers with the source row attached, then turns the answer into slides, sheets, or a report. Its pricing page lists Free at $0, Pro at $13.27 per month, Plus at $26.60, and Premium at $132.67, all billed annually.
How to decide in one pass
Gemini in Google Workspace and an upload-first agent answer different questions, so the choice is usually about inputs rather than features. Ask three questions before you pick a lane.
Where does the source material live? Inside Drive, Gmail, and Chat, the Workspace route is the shortest path. On your disk or in an export, an upload-first tool starts sooner.
Who has to trust the number? If a figure will be challenged in the meeting, choose the tool that can point back to the row it came from.
How often does this repeat? A monthly cycle rewards a saved prompt and a stable source list. A one-off analysis rewards whichever tool opens fastest.
Conclusion
The September 9 update moves Gemini from a helper inside one app to an assistant that produces a file while you stay put. For recurring internal updates built from Workspace material, that is a genuine time saving, and the preview-card and permission rules are sensible defaults.
Just be honest about your inputs. If most of your analysis begins with a file somebody sent you, you need a tool that starts there instead. Try Powerdrill Bloom on your next export and see the answer come back with its source attached.
Frequently asked questions
When did Gemini get cross-app task support in Google Workspace?
Google published the announcement on September 9, 2026. The gradual rollout to Rapid Release and Scheduled Release domains started on September 2, 2026. It can take up to 15 days for the feature to appear.
Which apps can I prompt Gemini from?
Google names Gmail, Drive, Docs, Slides, and Chat. You use the side panel in Gmail, Docs, and Slides, "Ask Gemini in Chat" inside Chat, and "Ask Gemini in Drive" inside Drive.
Can Gemini build a slide deck without opening Google Slides?
Yes. Google's own example prompt from Chat is "Create a deck outlining Project Zebra with the latest updates." The result is saved to Drive with a direct link to the presentation.
Does Gemini use my documents to train Google's models?
Google's announcement states that your data "is not reviewed by humans or used to train Gemini models." It adds that Gemini respects the authenticated user's existing access permissions and sharing policies.
What languages are supported at launch?
English only. Google writes that "at launch, this feature will be supported in English only," with support for more languages to be added later.