Changelog

What's new

• More natural podcast hosts. New podcasts use Gemini voices by default, for more natural, conversational delivery.
• Pick your voice model. A new Voice model section in the podcast creator lets you choose between Gemini and OpenAI voices. OpenAI voices still support personality styles like Energetic, Calm and Teacher.
• Ten new voices. Gemini comes with its own set of hosts, from Charon (informative) to Zephyr (bright), so you can find the right pair for your topic.
• A cleaner podcast creator. Style, Length and Focus are front and centre. Everything else is tucked away, one click from view.

What's new

• One summary across all your sources. On a dashboard with several sources, Generate Summary brings them together into one summary organised by theme, and cites each source it draws from.
• Clickable citations in summaries. Click a citation in a summary and it opens that source at the cited page. Page references like [Page 12] in AI chat are now clickable chips too.
• A quicker source picker. Click a source's name to use just that source. Tick its checkbox to add or remove it. The picker stays open while you choose.
• Text and Markdown files, natively. .txt and .md files upload faster and open in a clean text view.

What's new

• A Help menu in the sidebar. The ? button now opens Documentation, About and Get support, plus a "What's new?" feed of the latest updates.
• More room on mobile. A source opens with more space for the panel below it, so the Create button is right there. The divider between them is easier to find and drag.
• Tidier menus. Dropdown menus across the app share a cleaner, rounded style.

What's new

• A real icon library. The page-icon picker now has a searchable set of 1,500+ filled icons, grouped by category — books, science, arrows, brands, and more. Search by name or keyword ("favourite" finds the star, "delete" finds the bin).
• Filled, not outlined. The new icons are solid-filled for a cleaner, more polished look on your pages, sidebar tabs, and cards.
• Works everywhere. Whatever icon you pick shows up consistently — in the header, the sidebar, breadcrumbs, search, recent pages, and source lists.
• Your existing icons are untouched. Any icon you've already set stays exactly as it was.

What's new

• @-mention a source and it just works. When you @-mention a specific source in AI chat, the assistant now searches that source for its answer — no more "I can't read that" or errors.
• Cleaner AI formatting. Checklists and tick-mark lists the AI writes now render as proper lists instead of a run-on paragraph.
• Rename a source without renaming the page. Renaming the first file in a source list no longer changes your whole dashboard's name — they're independent now, and the AI's source pickers always show the right, up-to-date name.
• Tidier source menu. The source ⋯ menu matches the rest of the app, with a "Created by" detail and cleaner spacing.

What's new

• Published pages honour wide mode. If a page is set to wide, its public/shared version now uses the same full-width layout instead of the narrow default.
• Cleaner public layout. Fixed the spacing above the title so it isn't cramped, and tidied up the public view.
• Due dates show up on your home dashboard. Adding a Due Date to a page now surfaces it on the home calendar and the "Due Soon" list.

What's new

• Lock pages from the ⋯ menu. Workspace admins can now lock any page — reference material, a shared study doc, an answer key — so members can read it but can't change it.
• View-only for everyone but admins. When a page is locked, members (and shared editors) drop to view-only and can't comment. Admins can still edit freely.
• Live for everyone. Lock or unlock a page and everyone viewing it updates instantly — no refresh needed. A lock icon shows on the page's sidebar tab so it's clear at a glance.

What's new

• Members can build quizzes, not just admins — everyone in a workspace can create and edit them.
• A smoother builder. The add-question bar stays put as you scroll, and the view jumps to each new question as you add it.
• Fairer generated quizzes. Auto-generated multiple-choice questions now randomize where the correct answer lands instead of favouring one position.

What's new

• One switch to turn AI off. Workspace admins can disable AI for members — useful for exam settings, younger cohorts, or workspaces that just don't want it.
• The app adapts, it doesn't just grey things out. With AI off, the assistant panel, "chat with this" actions, quiz/flashcard auto-generation, and canvas AI nodes all step aside cleanly instead of showing dead buttons.
• Members still get everything else. Pages, boards, canvas, quizzes, and chat all work as normal — only the AI-powered extras are hidden.

What's new

• iPhone photos just work. HEIC images are converted on the fly, with a preview before you send — no more "Upload failed" when sending a photo straight from your camera roll.
• Cleaner multi-image grids. Send several photos at once and portraits keep their shape instead of getting cropped awkwardly.
• Fewer send failures. Image-only messages, mid-conversion sends, and a handful of edge cases that used to silently drop are fixed.

What's new

• Read-only channels. Workspace admins can lock a channel or DM so members can read it but can't post — good for announcements or archived conversations.
• Clear at a glance. A locked channel shows a lock icon in place of the # in the sidebar.
• Instant for everyone. Locking and unlocking, and admin promotions, now update live for everyone in the workspace without a refresh.

What's new

• Click a citation, land on the passage. When the assistant cites a source, clicking the citation now opens the PDF and highlights the exact sentence it used — not just the page.
• Cleaner citation tooltips that show the real source name.
• Multi-source answers now navigate to the correct document every time.

What's new

• Page emojis come across. Imported Notion pages keep their emoji icons.
• Cover images become page backgrounds. A page's Notion cover is carried over as its background.
• Bookmarks import as link previews — proper preview cards, not bare images.
• Fixed access and titles so imported pages open correctly and no longer show up as "Untitled".

What's new

• Layering controls. Bring canvas items to front or back, and step them forward or backward — with hover and click landing on what you actually see.
• Smarter snapping. Alignment guides prefer centre-to-centre and use each item's real size, so things line up where you expect.
• Image cropping and rotation. Crop from the edges and rotate items individually; images get rounded corners.
• Present mode matches the canvas — what you design is what you present, starting from whatever you've selected.

If you study anything with equations, chat answers now read the way they should. Formulas are typeset properly instead of appearing as raw LaTeX.

What's new

• Equations render inline and as blocks. Fractions, integrals, sums, Greek letters, and the rest display cleanly wherever the AI uses them.
• More reliable across the board. Math renders consistently whether it's a quick inline term or a full derivation, including in summaries and the canvas AI chat, not just the main chat.
• Prices stay prices. A dollar amount like $10 no longer gets mistaken for the start of a formula.

Nothing to turn on. Ask a maths question and the answer comes back properly formatted.

Paper can now build slide decks. Point it at a document, video, or PDF and it drafts a full presentation (titles, bullets, and layout) that you can edit however you like.

What you can do now

• Generate a deck from your source. Open the Slides feature and Paper turns your material into a structured presentation with real typographic hierarchy, not just walls of text.
• Edit on a real canvas. Each slide is a frame you can open and edit directly: move text, resize elements, drop in images, and rearrange, with the same freeform editing you get everywhere else in Paper.
• Pick a theme. Choose from a set of designed themes (fonts, colors, accents) and the whole deck restyles instantly.
• Remix a slide. Not happy with one slide? Regenerate just that frame from a prompt or a different layout, without touching the rest of the deck.
• Present. Play the deck full-screen in order, straight from Paper.

Built on your workspace

Decks live alongside your other pages, so they're searchable, shareable with your workspace, and collaborative like everything else. Generate once and your whole team can open, present, or keep editing.

Canvas pages got a big editing upgrade. Everything is more direct, more precise, and closer to the tools you already know.

What's new

• Shapes. Add rectangles, ellipses, diamonds, and stars (with adjustable points). Restyle fill, stroke, corner radius, and opacity from a floating toolbar.
• Rich text nodes. Full text styling (font, size, weight, color) plus markdown-style lists that continue on Enter. Add links inline.
• Sticky notes and text now drag from anywhere. Grab a node anywhere on its body to move it; double-click to edit. No more hunting for the top edge.
• Image cropping, Figma-style. Double-click an image to reveal the full frame, drag to reposition, and resize the crop by its corners. Press Enter to commit.
• Remove image backgrounds in one click. It runs right in your browser, no extra step.
• Search Unsplash without leaving the canvas to drop in a photo.
• Group and multi-select. Shift-click multiple nodes for a single bounding box, then right-click to group (or Cmd/Ctrl+G). Duplicate whole groups at once.
• Alignment snapping. Nodes snap to each other's edges and centers as you drag, with equidistance guides for clean spacing.

Cleaner selection

Selection now uses a consistent blue outline with crisp square handles and a live dimension label, plus real-time cursors so you can see where teammates are working.

On video and PDF pages you've got two things side by side: the source on one side, the AI panel on the other. Now you can give either one the whole width.

What's new

• Maximize the AI panel to read a long summary or chat without the video or PDF taking up half the screen.
• Maximize the source to focus on the document or watch the video full-width.
• Restore the split anytime. A subtle control on the edge brings the other side back.
• Your choice sticks. The layout you pick is remembered, so pages open the way you left them.

The collapse controls stay out of the way. They appear on the divider between the panels and never cover your toolbars.

If you use Paper with a workspace, the AI features are now collaborative. Every chat, summary, flashcard deck, quiz, and podcast has a share toggle.

What you can do now

• Share anything AI generates. Click the pill next to the title of a chat, summary, deck, quiz, or podcast. Teammates see it immediately.
• Continue each other's chats. Anyone can add follow-up questions or refine an answer. Names show on the messages people sent.
• Watch live. When a teammate is asking a question in a shared chat, you see the response as it's written.
• Save AI credits across your team. Generate a summary or podcast once, and everyone in the workspace can read it without regenerating.
• Hide what you don't want to see. Use "Hide for me" in the kebab menu to remove a shared item from your list without affecting anyone else. A "Show hidden" link at the bottom lets you restore.

Privacy by default

Nothing is shared unless you explicitly turn on the share toggle. Existing chats and summaries stay private. Only the creator can unshare - teammates can hide-for-themselves but can't take away everyone else's access.

Sharing is scoped to your workspace. A shared chat on a private document isn't visible outside your workspace, even if you later publish the document publicly.

What's new

• Publish any page or canvas. Flip a switch to get a public URL you can share. No login required for viewers.
• Cascade publish for canvases. When you publish a canvas with embedded pages, a confirmation modal lets you publish those too so viewers see the full picture.
• Read-only really means read-only. Public viewers can't edit or drag things in - the shared view feels finished.
• Resize still works on canvases so viewers can adjust node sizes for their screen without saving changes.
• PDFs work standalone at their public URL, not just embedded in a canvas.

What's new

• Timeline view for boards. See tasks laid out over time with drag-to-reorder.
• Task dependencies. Link tasks with arrows; the critical path is highlighted so you know what's blocking progress.
• Files tab gets a proper uploader, date filters, and a redesigned row.
• Linked dashboards propagate live: property, title, and reorder changes sync across cards in real time.

What's new

• .docx and Office file uploads are more reliable when several people upload at once.
• Large PDFs handle better during peak load - the compression step is faster and uses less overhead.
• Older AI chats that had gone quiet are back. If you had a chat on a document from earlier this year that stopped responding, it should work again.

What's new

• Public workspace channels with a discovery view. Browse, preview, and join without needing an invite.
• Live presence. Every dashboard shows who else is currently viewing it.
• Huddles. Start an audio/video call from any dashboard header. Set your mute, camera, and background before you enter.
• Screen share plus minimise, hide-stage, and mobile-friendly controls.
• @-mentions in pages with polished popover.

What's new

• Length up to 90 minutes. Set a target duration and get a script paced to fill it.
• Two hosts with distinct personalities. Pick voices per host and give each one direction ("speak with enthusiasm", "like a professor").
• Any language. Generate in French, Spanish, Japanese, or set a default language for all your podcasts.
• Multi-source. Combine several documents into one podcast that draws from all of them.
• Synced transcript that highlights the current line as it's read.

What's new

• Paste from Notion. Headings, lists, callouts, code, and images all come through as native Paper blocks.
• Real tables with cell drag-swap and column-width sizing.
• Callouts and accordions that persist their content and expand smoothly.
• Cross-context drag and drop. Move blocks between the editor, board view, and linked-dashboard views.
• Live collaboration. See teammates' cursors and edits in real time.
• Mobile. Drag handles, block reorder, and toolbars work on touch.
• Table of contents for long pages.

What's new

• Section-level detail for long documents. You get a hierarchical summary that mirrors the document's actual flow, not a single dense paragraph.
• Works on any document. Paper detects chapters, numbered headings, or infers natural breaks for documents with no explicit structure.
• Consistent depth. Every section gets similar treatment, so the summary doesn't skew toward whatever came first.
• Faster overview answers. Asking "what's this document about?" in chat now uses the cached section summaries, so you get an answer in a couple of seconds instead of twenty.

What's new

• Broad summary questions get a short high-level overview, not a chapter-by-chapter inventory.
• Section-specific questions ("what does Section 1.3 say?") answer strictly from that section, with related sections suggested at the end.
• Topic-learning questions ("explain X to me") build the concept in the order the document introduces it.
• Fact-lookup questions give you the answer in the first sentence.
• Better citations. Every claim carries a page reference, and multi-source citations include the source name so you jump to the right document.

What's new

• Big PDFs work. 500-page textbooks and 1000-page reference documents upload and become chatable in a few minutes.
• Scanned PDFs are readable. Pages that used to be images now get OCR'd automatically. Highlight, copy, and search work as if the document were born-digital.
• Smart handling for mixed pages. Pages that are mostly text but have a diagram or figure only OCR the parts that need it.
• Visible progress. Long extractions rotate through status messages so you can tell Paper is working, not stuck.