Back to Blog
•14 minutes read

Scholarly May 2026 Updates Recap

May 2026 turned folders into a full study workspace, overhauled AI Video Lectures with a textbook-quality visual system and Video Modes, added Claude to AI Chat, brought Response Variants, Captions, Interactive Video Questions, Chat Canvas, a sharing overhaul, AI Slides themes, and dozens of mobile and reliability improvements.

By ScholarlyUpdates
Share:

May 2026 may be our biggest month yet. Folders went from being a place to file things to being a full workspace you can study in. AI Video Lectures got a complete visual rebuild plus eight new teaching styles. Claude joined AI Chat. We added Response Variants, Captions, Interactive Video Questions, Chat Canvas, and a top-to-bottom sharing overhaul. Plus a brand new Study Stacks tool, a Cheat Sheet Maker, an AI Cue Cards tool, an Anki-ready .apkg export, and a long list of mobile and reliability improvements. Here's the full rundown.

Major Features

Folders Are Now a Workspace

Every folder is now a complete study workspace, not just a place to file things. Open one and you'll see:

  • Sources panel on the left — every file in the folder, including anything in subfolders.
  • Folder-scoped chat in the middle — ask about "this folder" or "these sources" and answers stay grounded in the folder.
  • Create with AI panel on the right — make a podcast, video lecture, slide deck, flashcards, or research session inside the folder, and the result lands right back in the folder automatically.

A few details that make it click:

  • Drag and drop files anywhere in the folder workspace to upload straight into it.
  • The Create modal source picker auto-scopes to the folder so picking inputs is one click instead of three.
  • Pending items show up live in the sources list while they generate.
  • When AI Chat creates something for you in a folder chat — a podcast, deck, slides — it's added to the folder.

How to use it: Open any folder from your sidebar or home page. Click into the Create with AI panel on the right to make something new, or drag files in to upload. Anything you build stays in that folder.

AI Video Lectures — Complete Visual Rebuild

The biggest single quality jump we've shipped for video. AI Video Lectures now look and feel like a beautifully designed textbook in motion:

  • Warm paper backgrounds, serif headlines, and clean diagram-style scenes.
  • Smarter scene types that the AI picks automatically — definitions, comparisons, step-by-step processes, equations with real math typography, count-up stat animations, charts, and key-insight callouts.
  • For STEM topics, lectures now lean on real animated math and diagram clips — derivations, plots, geometric figures, and mechanisms come to life on screen instead of being shown as static illustrations. Closer to a 3Blue1Brown explainer.
  • Same narration, same speed, same 27+ languages — just dramatically better-looking output.

We also added a render quality check: if a video is missing audio, ends early, or has a silent narration track, we catch it and regenerate automatically instead of shipping a broken file.

Video Modes

You can now choose the teaching style for any AI Video Lecture. Pick from eight modes:

  • Standard — clean explainer with structured visuals.
  • TL;DR — fast-pace overview hitting only the essentials.
  • Math — derivations, equations, and geometric figures.
  • Storybook — illustrated narrative for soft subjects and humanities.
  • Kids — friendly, simple visuals for early-learner-friendly explanations.
  • Debate — two-sided framing with point-counterpoint structure.
  • Case Study — concrete real-world example with analysis.
  • Podcast — conversational two-host style audio.

Standard and TL;DR are free on every plan. Paid plans unlock longer videos and more daily creations.

Video Lecture Length Tiers

We replaced the old Short/Medium/Long picker with three clearer tiers:

  • Quick Recap — now 1.5 minutes (up from 30 seconds), enough to actually teach a concept.
  • Standard — a tight, reliable 4–6 minutes.
  • Deep Dive — 6 to 12 minutes depending on the source. Pick this for exam-ready coverage.

If an overview comes out too short or doesn't cover enough of your material, it's now regenerated automatically instead of shipping cut short.

Interactive Video Questions

Every AI Video Lecture (and uploaded video) now prepares interactive questions from the transcript and pauses at key moments so you can check your understanding while watching. New this month:

  • Best and Last scores — see your top attempt and most recent attempt right next to the questions list, plus a full history.
  • Non-intrusive quiz card — questions now slide up as a small card at the bottom of the video. Hover or tap to expand and answer; the video keeps playing the whole time, so you stay in flow.
  • Toggle auto-pause on or off per video, and turn any checkpoint into flashcards or a quiz with one tap.

AI Slides Themes

AI Slides now come with visual themes. Pick a look for any generated deck:

  • Editorial — the classic look.
  • Vibrant — bold colors for engaging presentations.
  • Minimal Mono — clean monochrome for serious topics.
  • Pitch — designed for high-impact pitches.

Themes are free on every plan. Plus, every generated deck now runs a visual check on every page before finishing — catching overlapping text, clipped content, and broken layouts so the deck you download is clean on the first try.

Response Variants in AI Chat

You can now retry an AI response with a different model, keep each version available in the conversation, and switch between variants for follow-up messages. Got a Claude answer you weren't happy with? Retry with GPT 5.4 or Grok 4.3. Compare and pick the one you like — the conversation branches naturally from whichever variant you choose.

Chat Understands Your Library

Ask AI Chat to find one of your files, and it will. New library-search capabilities:

  • Search by name, type, or date: "what was that PDF I uploaded last week?" or "find my biology recording from Tuesday."
  • Tell Chat to actually read one of your files: "open the cardiology PDF and answer using it." It loads the PDF, recording, podcast, or video — plus the transcript for audio so it can answer faster.

AI Chat Manages Your Flashcard Decks

AI Chat can now add, update, and remove cards in the flashcard deck you're studying, with changes saved automatically. You can even edit cards by their position — "remove cards 16–22" — without needing to know card IDs.

Claude Sonnet 4.6 and Haiku 4.5 in AI Chat

Claude is now available in AI Chat. Pick Claude Sonnet 4.6 (now the default chat model for Plus and Ultimate) or Claude Haiku 4.5 for faster everyday replies. Both support thinking mode for tougher questions and work with everything else in chat: linked PDFs, recordings, web search, code, and file creation.

Captions for Podcasts and Video Lectures

AI Podcasts and AI Video Lectures now include captions. Toggle the CC button on the podcast player, or use your browser's caption controls on videos, to follow along while you listen. Supported across our 27+ languages.

Chat Canvas

You can now sketch ideas directly in chat. Draw a diagram, link it to a message, and ask AI about what you drew. Useful for working through math problems, explaining a concept visually to the assistant, or just thinking out loud on screen.

Sharing Got a Serious Upgrade

The Share window had a top-to-bottom redesign:

  • Copy link leads — the link bar and public/private toggle are right at the top, where you'd expect them. Email invites moved into a tucked-away section.
  • QR codes — generate and download a QR for any shared item. Great for showing on a slide or whiteboard so people can scan.
  • One-tap mobile share — on phones, the Share button hands off to your phone's native share sheet (Messages, WhatsApp, AirDrop, anything you have installed).
  • Public stats — when something you shared is public, see views, saves, and when it was last viewed.
  • "Clone" buttons are now "Save to my account" everywhere with a bookmark icon.
  • Sign-up prompts on shared links now name the specific item the visitor is looking at and lead with "Save this."

Request Access to Shared Links

If someone shared a private link with you — a flashcard deck, PDF, video, podcast, recording, research session, or chat — you can now ask the owner for access right on the page instead of hitting a dead end. The owner gets an email, can approve in one tap, and the content opens for you as soon as they do.

Smarter, Editable Memory

Scholarly's memory of you across chats is now manageable. Open Settings → Scholarly AI → Memories:

  • Add your own memories. Type any fact you want Scholarly to remember — your major, exam date, accommodations — and it'll inform future chats.
  • Mark anything outdated. Finished an exam, switched schools, changed your mind? Tap "Mark outdated" and that memory stops shaping future answers. Restore it any time.
  • Cleaner over time. When something about you changes mid-chat, the assistant now updates, supersedes, or retires the old memory instead of stacking near-duplicates.

Research File Deliverables

Deep Research can now create supporting files alongside a report:

  • Preview them right in the side panel.
  • Save generated PDFs directly to your library.
  • Redesigned side panel — reports and files sit together at a glance, sources tuck into a collapsible list.
  • Revise a file without creating duplicate cards; saved versions are kept when a deliverable is updated.

You can now also pick which AI model powers a Research report, so you can match the model to what you're researching.

New AI Models

Claude Sonnet 4.6 and Haiku 4.5 (May 11)

Both Claude models are available in AI Chat across paid plans, with thinking-mode support. Sonnet 4.6 is now the default new-chat model on Plus and Ultimate.

GPT 5.4 (May 12)

GPT 5.4 is now the recommended premium OpenAI chat model — a fast, capable all-rounder. GPT 5.4 Mini remains the free default.

Grok 4.3 with Instant Mode (May 1, May 21)

Grok 4.3 is available across AI Chat and AI Slides as a thinking model, with lower xAI pricing for faster, more affordable generations. The new Instant mode delivers fast replies when you don't need deep thinking.

Gemini 3.5 Flash and 3.1 Flash Lite (May 19)

Gemini 3.5 Flash replaces Gemini Pro as Google's high-quality chat model. AI Video Lectures now default to Gemini 3.1 Flash Lite for faster, more efficient lecture generation.

Kimi K2.6 with Thinking Traces (May 12)

Kimi K2.6 now shows its step-by-step reasoning in chat, just like our other thinking models — expand the "Thinking" section to see how it worked through your question.

New AI Study Tools

We expanded the AI Tools library significantly this month:

Slide and Presentation Tools

Study Tools

Flashcard Tools

We also refreshed the AI Study Tools page with all 14 tools in one cleaner grid and added a Study Tools and Resources hub grouped by what you're studying, with a use-case picker for med, nursing, law, CS, and high school.

Free vs Paid Plans

A few important plan changes this month:

  • Free accounts now get 3 daily AI creations that can be used for any combination of flashcards, podcasts, video lectures, slides, recordings, or deck edits. The daily budget is now unified — you'll know exactly where you stand.
  • All AI Video Lecture modes are free on every plan. Free users can pick any teaching style (Standard, TL;DR, Math, Storybook, Kids, Debate, Case Study, Podcast); paid plans unlock longer Standard and Deep Dive runtimes.
  • AI Slides themes (Vibrant, Minimal Mono, Pitch) are free on every plan.
  • Plus and Ultimate tiers updated: Plus now unlocks longer AI Video Lectures, unlimited normal chat, 10 daily AI creations, 30 daily premium model messages including GPT 5.4, 1 hour of Voice Mode per day, and priority support. Ultimate now includes 100 daily AI creations and 300 daily premium model messages.
  • Free upload limit restored to 8 MB (including PDFs for flashcards).

Mobile — A Complete Pass

May was a huge month for mobile. Almost every surface in the app got a phone-specific pass:

Quick Create Button

A floating + button is now always within thumb reach on phones. Tap it to start a chat, upload a PDF, generate flashcards, record a lecture, or any other Scholarly action without opening the side menu first.

Flashcards on Phones

  • Swipe to navigate — flick left for the next card, right for the previous; flipping with a tap still works.
  • The flashcard editor stacks the question and answer on phones so each field is full-width.

Folder Workspace on Phones

On phones, folders open with Sources, Chat, and Create with AI each in their own tab so the Create panel isn't buried two viewports down. Folder workspaces also flip to the full 3-pane layout at iPad portrait (768px) instead of waiting for desktop. When a folder has content, it opens to Sources first so you immediately see what's inside.

Recording

  • Live recording keeps your screen awake for the whole capture so multi-hour sessions don't end early.
  • A floating Stop button stays visible while you record, so you can read the live transcript without losing your place.

AI Chat on Phones

  • Send, cancel, and voice buttons in chat are bigger — easier to hit, harder to mis-tap.
  • The Chat home screen now shows your recent chats as a stacked list so you can jump back into a conversation without opening the sidebar.
  • Tapping into the chat box no longer zooms the page on iPhone Safari.
  • Failed messages stay in the conversation with an inline Retry option instead of disappearing.

Other Mobile Wins

  • Move to folder from phone — tap the kebab menu on any item and pick "Move to folder."
  • PDF "Send to Chat" — appears when you select text by long-press on iOS and Android.
  • Slides fullscreen — visible close button on phones and swipe down to exit.
  • Podcast and recording scrub bars — fatter touch targets that actually land where you tap.
  • File previews in chat — now open as a full overlay instead of being crammed into a sidebar.
  • Share window — full screen on phones with bigger channel buttons and tappable Remove buttons on invites.
  • Settings — opens straight to the active section, with section tabs as a single scrolling row at the top.
  • Notch and Dynamic Island respect — toast notifications and the top app bar no longer get clipped on iPhone.
  • Voice mode microphone steps — correct, browser-specific instructions on Android and iPhone when access is blocked.
  • Welcome tour — opens full screen with larger buttons; we skip it entirely on phones where it didn't fit well.

We also fixed a long list of touch-only / hover-only bugs — pin, rename, archive, fold-into-folder, message actions, queue controls, file rename, recent files, in-page chat panels, video transcripts, PDF summaries, and flashcard highlights are now all reachable on phones and run to the true bottom of the screen on iPhone Safari.

Voice Mode Is More Reliable

Voice mode had a reliability pass this month:

  • Pre-permission explanation — the first time you start voice mode, we now explain what's about to happen before your browser asks for microphone access, so you know to click Allow.
  • Automatic reconnect — a brief disconnect no longer ends the session. Voice mode reconnects up to five times with progressive backoff.
  • Longer connection timeout — bumped from 10 to 20 seconds so slower networks can complete the handshake.
  • Clear failure messages — if voice mode can't start, you see why: blocked microphone, no microphone connected, or another app using your mic. The blocked-microphone case shows step-by-step instructions for Chrome, Safari, Edge, or Firefox.
  • No more mid-sentence cut-offs — we tuned how voice mode detects when you're done speaking.
  • iOS Safari resume — voice mode now resumes properly on iOS Safari, which sometimes started in a silent state.
  • Voice transcripts are no longer lost if saving them fails — they're kept safely and retried.

Connected Apps

We rebuilt connected apps so Google Workspace links Drive and Calendar together and gives Chat more reliable tools for finding files and events. Connecting Google with only some permissions granted now shows a clear error instead of silently enabling unconnected apps, and reconnecting after a temporary authorization issue retries gracefully instead of forcing a full re-link.

Flashcards

A bunch of flashcard improvements landed this month:

  • Anki exports now download as a direct .apkg deck file instead of a manual import text file.
  • Citations to source timestamps — flashcards generated from videos and recordings now link back to the exact moment in the source so you can jump straight to the lecture clip or recording timestamp.
  • Manual build tab in the Create Flashcards window — skip AI and write each question and answer yourself.
  • Fuller decks from large PDFs — paid plans now cover up to 1,000 pages.
  • Cleaner study view — lighter ghost styling for Hint / Favorite / Play Sound, the Teach Me button picks up the AI accent, the bottom controls space out evenly with bigger tap targets, and the session timer and keyboard shortcuts moved into the flashcard menu.
  • Quiz questions are noticeably more accurate — true/false sets are now balanced (no more runs that come out all false), and a correctly stated fact is marked true instead of being forced false.

How to Use the Highlights

A few quick walkthroughs:

Turn a folder into a study workspace

  1. Drag your study material (PDFs, recordings, videos, notes) into any folder, or upload them directly inside the folder.
  2. Use the folder chat in the middle panel to ask questions across everything in the folder.
  3. Click the Create with AI panel on the right to make a podcast, video lecture, slide deck, or flashcards from those sources. The result lands back in the folder automatically.

Try Video Modes

  1. Open Create AI Video Lecture.
  2. Pick a Video Mode — Standard, TL;DR, Math, Storybook, Kids, Debate, Case Study, or Podcast.
  3. Pick a length — Quick Recap (1.5 min), Standard (4–6 min), or Deep Dive (6–12 min).
  4. Add your source(s) and hit Create.

Retry an answer with a different model (Response Variants)

  1. In any AI Chat reply, click the retry icon.
  2. Pick a different model.
  3. Compare variants and switch between them for follow-up — your conversation branches from whichever variant you continue from.

Make AI Chat read one of your PDFs

  1. Type something like "open my cardiology PDF and answer using it" or "find my biology recording."
  2. Chat searches your library by name, type, or date.
  3. The file loads automatically (plus the transcript for audio) and Chat answers grounded in it.

Share with a QR code

  1. Open the Share window on any item.
  2. Scroll to the QR code section.
  3. Download the QR and drop it into a slide or whiteboard. Anyone who scans it opens the shared item.

New and Refreshed Help Guides

May was also a big month for the Help Center:

New Comparison Pages and Blog Guides

We shipped a fresh round of comparison and how-to content for students choosing between Scholarly and other tools:

Bug Fixes and Reliability

A long list this month — here's the user-visible highlights:

AI Chat

  • Failed AI Chat messages now stay visible with an inline Retry option instead of disappearing after a network drop or daily-limit hit — your text is preserved.
  • Hitting your daily Chat limit no longer stacks a toast and a popup; you see the upgrade window only, and your typed message is preserved.
  • AI Chat retries now use the original saved message and safely recover attached canvas context.
  • Failed Chat messages now recover automatically from brief mid-response hiccups instead of dropping your message.
  • Fixed scrolling on active conversations and a chat streaming issue that could stop some answers from finishing correctly.
  • Generated images in AI Chat now open full-screen when clicked.
  • Asking the AI to "teach me" or explain a flashcard now sends the actual card, so you get a real explanation right away.

AI Video Lectures

  • The Generate title with AI button on video pages now works (it previously failed with a "Not found" error).
  • Suspiciously short videos no longer trigger the share window or rate-this-video prompt and skip interactive checkpoints.
  • Errors now show the server's specific reason (for example, an unsupported source) instead of collapsing to a generic message.
  • AI Video Lectures that ran too long no longer get cut off without a clear timeout message.
  • Fixed an AI Video Lecture page that could keep loading interactive questions forever.
  • Quick Recap now hits its target length consistently.

AI Slides

  • Visual check on every generated page before finishing — catches overlapping text, clipped content, and broken layouts.
  • Clearer validation errors when Slides or Video Lectures fail — you now see exactly which field needs attention instead of "Invalid request data."
  • Fixed raw markdown styling sometimes appearing in generated AI Slides.
  • AI Slides now uses a cleaner light-mode editorial design for sharper output.

Uploads

  • Image and file uploads now retry automatically after a brief network hiccup instead of failing.
  • Uploading a batch where a few files are too large now continues with the ones that fit, instead of blocking the whole upload.
  • Files that became unavailable after picking them (moved, renamed, or synced by iCloud/Drive/OneDrive) now show a clear "please reselect" message.
  • Hitting your daily upload or file-size limit in the upload window now opens the upgrade option inline instead of showing a vanishing error toast.
  • Stopped a stale-build error from showing as a generic "upload failed" — the upload flow now correctly prompts you to refresh after an update.

Recordings and Podcasts

  • Podcasts now properly break content into multiple chapters instead of occasionally creating just one or two long sections.
  • AI Podcast audio no longer falsely shows a "taking too long" error during the first few seconds when bytes are still arriving.
  • New podcasts use smaller audio files for faster loading.
  • Recordings with missing audio metadata now still get transcribed instead of failing outright.
  • Recording transcripts that could fail completely when a single audio segment was malformed are now resilient.

Flashcards

  • Studying flashcards in flip mode now counts toward your study streak and stats (previously only the Questions/quiz mode did).
  • Flashcard creation failures now stay recoverable: if a source is too large, unreadable, or doesn't produce cards, the failed task stays visible and guides you to adjust the source instead of showing a dead-end retry.
  • Holding the next/previous key no longer rapidly skips cards.
  • Failed flashcard creation requests no longer count against your daily limit.
  • Fixed an issue where recently added flashcards could be lost when quickly switching between editing and studying.

Billing and Plans

  • Fixed the "accept this offer" button in the cancellation flow.
  • Simplified the Pricing page to a clear Free vs Ultimate comparison.
  • Switching to yearly billing now asks you to confirm first, so an accidental click won't switch your plan.
  • The upgrade window now shows specific copy for the action you tried, plus a Free column so you can see exactly what you're upgrading from.

Sign-in and Onboarding

  • Google One Tap sign-in now appears again on login, sign-up, and public pages.
  • Sign-in from a deep link (for example, a shared chat URL) now sends you to the original page instead of always dropping you on Home.
  • Public study tools now remember what you pasted, linked, or uploaded before sign-up and reopen the right create flow after you finish joining.
  • Safari users no longer get stuck on a blank page after an app update — it refreshes automatically.
  • After we ship an update, the app now refreshes itself instead of occasionally showing a load error.

Light Mode Polish We did a top-to-bottom sweep of light-mode contrast across the dashboard — PDF, recording, podcast, video, research, content, folder, and settings pages now have properly dark text, clearer card backgrounds, no missing borders, and no stray bright-white panels.

Performance

  • Background music in AI Video Lectures is now much softer — a faint bed under the narration instead of competing with the voice.
  • Faster public page loading across the homepage, blog, tools, and feature pages.
  • AI Chat responses are snappier in long conversations.
  • AI Video Lectures now generate much faster, and long lectures no longer time out before finishing.
  • Slides and video lectures can keep building while longer narration and render steps finish, so complex creations feel less stuck.
  • PDF highlights and summaries now finish faster.

Looking Forward

May 2026 turned Scholarly into a genuine end-to-end study workspace. Folders went from filing cabinets to fully functional workspaces. Video Lectures got a complete creative rebuild plus eight teaching styles. AI Chat learned to manage your library, edit your flashcards, and switch between Claude, GPT, Grok, Gemini, and Kimi mid-conversation. Sharing, mobile, voice mode, and reliability all moved forward together.

If you have any questions or feedback, reach out at hello@scholarly.so.

Try Our Popular AI Study Tools

Transform your study materials into interactive learning experiences with our most popular AI-powered tools:

PDF to Flashcards

Convert lecture notes and textbooks into study flashcards instantly

Text to Flashcards

Turn any text or notes into comprehensive flashcard sets

Image to Flashcards

Convert diagrams and handwritten notes into digital flashcards

YouTube to Flashcards

Generate flashcards from educational video content

View All Study Tools

Keep Reading

Scholarly now supports Claude

Scholarly now supports Claude

·Updates
Scholarly April 2026 Updates Recap

Scholarly April 2026 Updates Recap

·Updates
Introducing AI Video Lectures: Turn Your PDFs into Animated Study Videos

Introducing AI Video Lectures: Turn Your PDFs into Animated Study Videos

·Updates
Scholarly February 2026 Updates Recap

Scholarly February 2026 Updates Recap

·Updates

© 2026 Scholarly

English · 日本語

Company

hello@scholarly.so