Claude Slides: What's New, How to Access It, and Alternatives (2026)
On September 16, 2026, Anthropic folded Claude Cowork back into the main Claude app. The same announcement introduced two new output surfaces: Claude Docs and Claude Slides. Both live inside an ordinary conversation now, so you no longer pick a place before you start a task.
This page covers what shipped, who can reach it today, and what else produces a deck from the data you already have.
What Anthropic announced
The announcement is titled Claude Cowork and chat are now one Claude. Its structured date is September 16, 2026.
The core change is a merge. In Anthropic's words, "Claude Cowork and chat are merging into one Claude." Cowork had been a separate space for longer work, and Claude Design a separate space for visual work.
Anthropic explains the reason directly: "People used both, and told us the frustrating part was deciding where a task belonged." The company adds that "What they'd started in one also didn't carry into the other."
So the fix removes the choice. Any conversation can now reach what Cowork and Design used to do.
What Claude Slides actually does
Claude Slides is one of three surfaces that shipped together. The announcement puts it plainly: "Claude Docs and Claude Slides are new today, and Claude Design now works inside your conversations too."
The described loop is short. "Ask for a presentation, and Claude drafts the slides."
From there you get three exits. The page lists them as one sentence: "You can edit directly, present straight from Claude, or download as PowerPoint or PDF."
Two more details matter for anyone who shares decks. First, "Anything you make with Claude Design, Slides, or Docs lives at one shareable link you can open on your phone." Second, editing is not limited to prompting — "You can select an element and move it, or tell Claude what you want changed."
The report-and-deck pairing
The worked example in the announcement is the part worth reading twice. A weekly report is due at noon. You ask what moved in the pipeline, then add a request for "the highlights in five slides for the leadership meeting."
What comes back is a doc and a deck. The announcement notes why they agree with each other: "Both came out of the same conversation, so the slides already match the report."
That is the real claim here. Not that a deck appears, but that the deck and the document behind it share one source.
What Claude Docs adds alongside it
Claude Slides did not ship alone. Claude Docs arrived on the same day. The announcement describes a collaborative loop for it: "Ask for a document, and you and Claude write it together."
That pairing is deliberate. The worked example in the post produces both artifacts from one request, and the report comes back "as a doc with notes from your teammate."
Claude Design is the third piece. It existed before as its own surface, and it keeps working that way. The announcement is explicit on this point: "If you use Claude Design on its own, it keeps working as before." What changed is that it now also runs inside conversations.
For teams, the practical effect is fewer handoffs. A comment left on a slide goes to the same place that produced the slide. The announcement describes exactly that motion: "You fix a line yourself, leave a comment for Claude on a slide, and share it."
Who can use it today
Availability is staged, and the announcement is specific about the order.
| Plan | Status as described on the announcement page |
|---|---|
| Pro and Max | Rolling out first, "over the next few weeks" |
| Team and Free | "will follow soon" |
| Enterprise | Admins "will hear from us at least 30 days before anything changes" |
Two constraints sit alongside that table. All three surfaces are "in beta on paid plans," and "Enterprise admins choose when to turn them on."
The rollout reaches "the Claude app on web, desktop, and mobile." For people already on Pro or Max, the page says "There's nothing to turn on."
If you were a Cowork user, nothing was moved. The announcement lists what carries over: "your chats, projects, artifacts, connectors, and skills."
What the announcement does not cover
The page describes no separate price for Claude Slides. It also gives no general availability date and no seat tier of its own.
Those are not limits — they are simply topics the announcement does not address. Anyone building a budget around this should wait for the pricing page rather than infer a number. The full capability list is pointed to the Help Center rather than spelled out in the post.
How the approval model works
One design decision here affects how much you can hand over. By default, the announcement says, "Claude asks before taking an action."
You can change that. The page describes an option to "let it keep working and check in only when something needs a closer look." Anthropic closes the paragraph with a short line: "You keep the final say."
Scheduling is part of the same picture. The example ends with a recurring task — "Schedule it for every Monday, and Claude starts on the report without being asked."
Alternatives that build a deck from your data
Claude Slides starts from a conversation. If your starting point is instead a spreadsheet, an export, or a stack of PDFs, several other routes produce the same artifact.
| Route | Where the content comes from | Exports to PowerPoint |
|---|---|---|
| Claude Slides | The conversation you are already in | Yes, per the announcement |
| Powerdrill Bloom | Files you upload, plus connected open data | Yes |
| Google Slides with Gemini | Text prompts inside Workspace | Native Google format |
| Manual build in PowerPoint | Whatever you paste in | Native |
Powerdrill Bloom works from the file rather than the chat. You upload the Excel or CSV, ask for the analysis, and the deck is generated from the numbers that came back. It also runs Claude Skills, so a workflow you already wrote can drive the build. The walkthrough for that sits in converting a topic to a PPT with Claude Skills.
If the input is a spreadsheet, auto-generating PPT slides from Excel files covers that path directly. For a shorter version of the same job, there is making a presentation from Excel data in minutes.
For background on the product that just got merged, see what Claude Cowork was. That page describes the standalone surface, which is the thing this announcement changes.
Ready to build a deck straight from your own numbers? Start with Powerdrill Bloom.
Where the conversation-first approach fits
Claude Slides assumes the material is already in the thread. That assumption holds well for work that started as a discussion — a strategy memo, a project recap, a set of decisions taken over a week.
It fits less naturally when the source of truth is a file nobody has opened yet. A quarterly export, a survey download, a ledger pulled from an accounting system: none of that is in a conversation until someone puts it there.
Both shapes are legitimate. The question is only which one your week actually looks like.
| Your starting point | Better fit |
|---|---|
| A thread where the thinking already happened | Conversation-first, such as Claude Slides |
| A raw file nobody has read yet | File-first, such as Powerdrill Bloom |
| A recurring report with fixed inputs | Either, if it can be scheduled |
The scheduling note in the announcement matters for the third row. A deck that regenerates every Monday is only useful if the inputs refresh too.
What this signals about deck generation
Three product moves in 2026 point the same direction. Decks are becoming an output of the thing that did the analysis, rather than a separate step afterward.
The merge is the clearest version of that yet. A report and a deck generated from one conversation cannot disagree with each other, because there is only one set of numbers underneath.
The open question is provenance. When a slide says revenue fell nine percent, a reader eventually asks where that came from. Tools that keep the source file attached to the answer make that question cheap to answer.
Claude Slides answers it through the conversation. Everything on a slide traces back to something said in the thread, which is auditable as long as the thread is.
File-first tools answer it differently. The number traces to a cell in a workbook you can open. Neither is stronger in the abstract — they just fail in different places, and it is worth knowing which failure your reviewers will catch.
Trying it on a real task
The announcement suggests a way in that avoids a toy test. Its advice to existing chat users is short: "try the next report or deck you'd normally build yourself."
That is better guidance than it looks. A sample prompt tells you the surface renders slides. A real report tells you whether the slides survive contact with your own numbers and your own reviewers.
Three prompts from the post are worth reusing as written, because they show the expected shape of a request. One asks Claude to write a recurring report the way it is always written, flag anything that slipped, and put the highlights in five slides. Another requests progress checks from a phone while the work continues.
The third is about tone rather than content. The announcement notes that "If something's unclear, Claude asks" — so an ambiguous request produces a question, not a guess.
Start with a deck where you already know the right answer. You will spot a wrong chart faster in material you have presented before than in something new.
A short checklist before you present from it
Claude Slides is in beta, and a beta deck still has to survive a real room. Four checks cover most of what goes wrong.
Confirm the numbers against the source. A deck generated from a conversation inherits whatever was said in that conversation, including anything stated loosely earlier in the thread.
Check the export before the meeting, not during it. The announcement describes PowerPoint and PDF downloads, and fonts sometimes shift between an in-app view and a downloaded file.
Decide who owns the shareable link. One link for a deck is convenient, and it also means access is managed in one place rather than per attachment.
Confirm your admin has turned it on. For Enterprise, the announcement is clear that admins choose when these surfaces become available.
| Check | Why it matters for a beta surface |
|---|---|
| Numbers traced to a source | The thread is the input, so loose phrasing carries through |
| Export opened once, early | Layout can differ between the app and a downloaded file |
| Link ownership agreed | Sharing is per link rather than per attachment |
| Admin toggle confirmed | Enterprise availability is an admin decision |
Frequently asked questions
What is Claude Slides?
Claude Slides is a presentation surface that Anthropic shipped on September 16, 2026. It drafts slides inside a normal Claude conversation. The announcement describes editing in place, presenting from Claude, and downloading as PowerPoint or PDF.
How do I access Claude Slides?
It is rolling out to Pro and Max plans first, across web, desktop, and mobile. Anthropic says there is nothing to turn on for those plans. Team and Free are described as following soon, and Enterprise admins get at least 30 days of notice.
Is Claude Slides free?
The announcement places all three new surfaces "in beta on paid plans." It describes no separate price for Claude Slides itself. For anything more specific, the pricing page is the source to check rather than this post.
Can Claude Slides export to PowerPoint?
Yes. The announcement states you can "download as PowerPoint or PDF." It also describes a single shareable link for anything made with Claude Design, Slides, or Docs.
What happened to Claude Cowork?
It merged into the main Claude app. Anthropic says existing work carries over, listing chats, projects, artifacts, connectors, and skills. What Cowork could do is now reachable from any conversation.
Sources: Anthropic, "Claude Cowork and chat are now one Claude," claude.com/blog, September 16, 2026.