Ask an AI assistant to restyle a forty-slide deck and you find the ceiling fast. It tells you what to change. You make the changes.
Two releases in the last week moved that ceiling. On October 30, 2026, Gemini in Google Slides began editing multiple slides or an entire presentation from a single prompt. A day earlier, the Gemini app started creating, previewing and editing files inside the conversation. AI is moving from answering questions to producing the artifacts your team ships.
What Google shipped
Gemini in Google Slides, October 30. From the Slides side panel, one prompt updates content across a deck, refreshes visual styles in bulk, or matches the style of another presentation. Gemini works out which slides need to change and applies edits directly to your native, editable slides. Google's examples:
- "Update this whole deck to match the style of this [reference file]"
- "Update this presentation to include more details based on [reference file]"
- "Summarize slides 5 to 7 into a single slide"
When it finishes, a summary of the changes appears in the side panel. Edits land as if you made them, so undo and version history both work. The feature is English-only at launch, and Workspace customers keep higher multi-slide creation limits through at least December 1, 2026.
Gemini app file creation and editing, October 29. You can now generate, preview and edit files inside the chat. Expand to full screen to type edits directly, highlight text to ask a question, or have Gemini refine it. Google lists three:
- Slide decks. Draft an outline or slide content, preview the layout in chat, adjust tone and structure, then open in Google Slides or download as PPTX or PDF.
- Spreadsheets. Generate trackers, tables and data models from a prompt, preview them, then open in Google Sheets.
- Documents and PDFs. Draft Docs, PDFs, LaTeX, Markdown or TXT files, then upload to Drive.
PDF preview and editing works on web and mobile. Documents, spreadsheets, decks, Markdown, TXT and code are web-only.
The wider cluster
These two did not land alone. Here is the same window.
| Date | Release | What it removes |
|---|---|---|
| Oct 30 | Edit entire presentations with Gemini in Slides | Manual restyling across a deck |
| Oct 29 | Create, preview and edit files in the Gemini app | The trip to a separate editor |
| Oct 28 | Immersive explorations in the Gemini app | Static explanations of layered topics |
| Oct 28 | Fill with Gemini cell-level web citations | Unverifiable generated spreadsheet data |
| Oct 23 | Publish interactive Gemini apps to Google Sites | Handing an internal tool to engineering |
| Oct 23 | Analyze BigQuery data in natural language in Sheets | Writing SQL before you can ask a question |
The direction is consistent: each release removes a handoff between thinking and producing. Fill with Gemini shows the substance. Cell-level web source citations now attach to each cell, capacity runs to 3,000 cells per request, and Google's study of 95 participants measured up to 55 times faster than manual entry on a 500-cell task.
The pattern: from answers to artifacts
Chat was the first interface. You asked, you read, you acted, and the artifacts were still yours to build.
The change is that the model now holds the artifact. It produces it, takes an edit instruction against it, and hands back a native file you keep working in. The output is a document to review, not a message to read.
CommandLyne already works this way
The same shape has been the core of CommandLyne's canvas since we built it. Canvas is a side workspace (opens in a new tab) for structured and interactive content beyond a normal chat response. Open a chat with an agent, look at the right-side panel, and click the Canvas tab. You get a live artifact beside the conversation, covering flowcharts, architecture diagrams, data visualizations, UI mockups and presentation decks. Two prompts we use ourselves:
- "Generate a flow diagram for user onboarding."
- "Visualize the system architecture we discussed."
Artifacts are not one-shot. Ask for a change and it updates in place while the conversation stays alongside it. Version history holds each revision, so a change you dislike is one step back, not a rebuild.
Files in, files out. The plus menu takes PDF, Word, Excel and text files alongside images, voice and video. Once a document is in, ask the agent to summarize it, extract data, generate a report, or convert the content into tables. Large files run as background tasks, so a long analysis does not hold the conversation.
One instruction, many systems. This is where the shape goes beyond a single document. The CommandLyne Assistant (opens in a new tab) decomposes a plain-language request into an ordered, multi-step plan across the tools you already use: Gmail, Calendar, Drive, Docs, Jira and web sources. One instruction can retrieve context, produce a deliverable and send it, without moving between tabs.
Every artifact stays governed. Model choice routes per task through multi-model orchestration (opens in a new tab), so a restyling pass and a hard analysis need not run on the same model. Enterprise memory keeps your formats and working patterns. Every action is scoped to your permissions, and you get Preview, Approve, Adjust, Cancel and Rollback before anything changes.
| Gemini in Slides and the Gemini app | CommandLyne canvas | |
|---|---|---|
| Where the artifact lives | Inside Google Slides, Docs and Sheets | In a side workspace beside the chat |
| What it produces | Decks, spreadsheets, docs, PDFs, code | Flowcharts, architecture diagrams, data visualizations, UI mockups, decks |
| Inputs it accepts | Drive files and the prompt | Drive files, PDF, Word, Excel, text, images, voice, video |
| Systems in one request | Google Workspace | Google Workspace, Jira, Confluence, Notion, Slack, Salesforce |
| Before execution | Undo and version history | Preview, Approve, Adjust, Cancel, Rollback |
What to reach for, and when
Reach for Gemini in Slides when the deck already lives there and the change is a restyle. It edits your native slides, so collaborators see the same file.
Reach for the Gemini app for a first draft of a document, sheet or deck without opening anything else.
Reach for CommandLyne when the work crosses systems. A quarterly review that pulls from Jira, reads a PDF from Drive, produces the deck and posts it to a channel is one instruction, with the model routed per step and every action previewed before it runs.
The assistant that produces the artifact beats the assistant that describes it.
Try CommandLyne to build artifacts beside the conversation with governance on every step, or book a demo and we will walk through it with your stack.
Try CommandLyne (opens in a new tab) | Book a demo (opens in a new tab)

