Dev Log

New features, improvements, and bug fixes. A timeline of updates, improvements, and new features added to CoreWise.

Have questions? Email support@corewise.video

2026-07-08

• Grok extractions now run on a faster, smarter default model

2026-07-07

• Raised the monthly extraction limits on every tier

2026-07-05

• Instagram image and carousel posts can now be extracted from their captions

2026-07-04

• Public MCP server - connect an AI agent (Claude, ChatGPT, or any MCP client) to your CoreWise account using API keys from the new Profile "API & MCP Keys" card
• Pasting a YouTube link with stray spaces now works, and malformed video links fail fast with a clear message to re-copy the link
• Copy button added to code blocks in the docs

2026-07-03

• Fixed broken Arabic and Urdu synthesis translations

2026-07-01

• Improved transcript accuracy for videos with no public transcript
• Loading screen now explains the wait when a video has no captions and is being transcribed from audio
• Route loading now recovers faster after a new deploy

2026-06-30

• Wait-game audio now loads from the media CDN
• Swarm Bolt Storm fires less often

2026-06-29

• Super Debate / Debate+ now adds grounded rebuttals, perspective coverage, citations, and a Roundtable loader
• Urdu language support added
• PDF export now renders Cyrillic, Greek, extended Latin, Chinese, Japanese, and Korean more reliably
• Language picker now places Urdu next to Hindi
• PDF export now matches the language you're viewing the extraction in

2026-06-26

• Improved reliability for very long transcripts

2026-06-24

• Chat Projects are now unlimited on the Core tier
• Jet Strike on mobile - zoomed in so sprites and HUD read bigger, taller portrait field, control hints hidden on touch
• Added a Screen shake toggle to /play settings

2026-06-21

• New Swarm boss: The Riptide - a tide-serpent that drags you with undertow
• Swarm: 10 new pacts with per-pact limits and a wider magnet range
• 6 new Swarm medals and a new Jet Strike medal - daily play cap raised to 17
• Chat Projects panel redesign - cleaner empty states, right-click menus, grounding note in the header
• Swarm Warden rework - erupts swarmers across the arena and spills its brood on death
• Swarm Chrono Wraith hourglass slam now leaves a safe lane
• Swarm performance improvements during boss fights and hazards
• Fixed clipped aspect-ratio labels in the infographic studio

2026-06-20

• New Swarm boss: Grand Tilt - a multi-phase pinball boss with multiball and a tilt dash
• Two new Swarm elites: Harrow and Pylon
• Web browsing in Library quick-chat - use the globe to search the web
• Swarm boss and elite polish - Apex Drifter introduction, Event Horizon's field-wide slow, gentler boss HP scaling
• Chat Projects - ChatGPT-style grounded workspaces with nested threads, citations, and inline instructions

2026-06-19

• Swarm: 10 new milestone bosses - Storm Herald, Event Horizon, Mirage, Siege Warlord, Aegis Bastion, Tessellith, Loadstar Tyrant, The Reliquary, Leviathan of the Lull, and The Cartographer
• Swarm unique signature attack for every milestone boss
• Ultra Concise toggle on all Q&A surfaces - shorter, denser answers

2026-06-18

• Swarm Doomsday Clock and Headhunter boons reworked
• Swarm late-game elites - APEX mini-elites, Fanner, and Seeker

2026-06-16

• Council split-view composer aligned with synthesis panel
• Swarm pact damage scales boss HP so late-game bosses stay challenging

2026-06-15

• Swarm bosses arrive sooner and attack faster when you crowd them
• Swarm roomier 3D level-up and pact panels
• Warden boss mini-elite swarmers rot and burst into lingering pools

2026-06-14

• Jet Strike portrait and landscape layouts fill the screen edge-to-edge on mobile
• Swarm Tier-6 capstones - Last Stand, Avarice, and Berserker; Glutton and magnet passives reworked

2026-06-13

• Jet Strike on touch devices - playable from /play on phones and tablets
• Swarm 20 new arena backgrounds
• Improved mobile create menu sizing in the nav
• Bengali language support - full UI, extraction translation, and audio/podcast export
• Fixed missing Chat and Deep Research translations across existing languages

2026-06-12

• + create menu visible in tablet header up to 880px
• Swarm 3 new medals - Veteran, Stalwart, Shockwave
• Long co-host podcasts no longer cut off mid-dialogue
• Swarm Shard Halo and Bolt Storm evolved weapons tuned down

2026-06-11

• Fixed synthesis pages crashing on Safari versions before 16.4
• Improved Swarm performance during boss fights and overload effects
• /play dashboard arcade-row redesign - marquee hero, beveled gold pills; guest hero is attract-mode only
• Council copy-all and Markdown export for the full exchange
• Fixed Council default model lost on cold load
• Swarm crescendo tier-1 aura is rising embers
• Space Invaders - player shots pass through bunkers

2026-06-10

• Swarm bosses 9-11 - Pinball Titan, Chrono Wraith, Quicksilver Gemini
• Swarm dual-evolutions (ascended weapons) and post-boss pacts
• Swarm overload spectacle, power-fantasy crescendo, late elite affixes, late fodder floods
• Mobile - wider Deep Synthesis and extraction synthesis; touch-friendly hover actions
• Jet Strike rework - Wisdom Surge meter (Space), bomb on left-click, weapon tiers 4-5, wave-20+ event waves (Twin Bosses, Noise Storm, Gold Rush)
• Space Invaders expansion - destructible bunkers, pixel fleet, formation variety, Core beam escalation, CRT scanlines toggle, golden UFO, boss desperation finishers, run-stats scorecard, graze visibility, wave intro cards, combo callouts

2026-06-09

• Swarm bosses 6-8 - Brass Zodiac, Ossuary Seraph, Abyssal Ouroboros
• Swarm boss caches, level-up reroll/banish, enrage, nameplates, letterbox intros
• Swarm full pixel-art pass, run-stats card, in-game settings
• Clearer evolution hints on level-up cards
• Space Invaders overhaul - three new bosses (D-F), tiered boss pools, bonus stage, divers and elites, juice and polish pass, powerup pickup feedback, 25% more drops

2026-06-08

• Swarm Agents boon - gold ally drones hunt distant enemies
• Swarm passive tier-6 capstones across all passives
• Swarm auto-pauses when cursor leaves; resumes on hover (Jet Strike and Space Invaders too)
• Council sidebar - Older group collapsed, two-line session titles, layout fixes
• /play - gold gradient Start tile; leaner Swarm pause panel
• Swarm in-extraction overlay widened to 16:9 with larger playfield; XP gains 25% faster
• Medkits always heal to full in Swarm

2026-06-07

• Homepage extraction form - Customize visible before focus, Save-as-default bar, larger Synthesis and domain-focus cards
• Mobile nav — Discover and create (+) in the phone cluster
• CoreBot — tap opens Quick Access; swipe up opens bookmarks; first-visit hint on phones
• /chat on mobile — floating collapsed sidebar; swipe or tap outside to close; auto-close on conversation pick; model name hidden until picker opens; icon-only composer chips; tap-to-fill starter prompts
• Council added to nav create (+) menu
• Swarm 8 medals and 3 Rare score achievements (Ascended, Overdrive, Singularity)
• Mind maps — pinch/pan touch gestures on mobile
• Mind maps in Quick Access — auto-bookmark on open; deep links open directly
• Custom Image Studio — create Auto or Custom images from extractions (model picker, reference upload, Optimize prompt)
• Library personal archive split into Deep Synthesis, Audio, and Images tabs; Image Gallery for studio creations
• Deep Synthesis archive cards show cover art with View and Export dropdown
• "+" create menu in the nav — Image, Chat, Extract, Research
• Animated image launcher on Library extraction cards
• Journey achievement and Earn Points cards cleaned up for mobile
• Swarm Play again neon button; mines hurt bosses and elites more; first boss eased; Warden summons a 30-enemy rush

2026-06-06

• Swarm bosses arrive through a warp tunnel for the whole fight; richer animated game-over screen
• Space Invaders bosses get a warp-in entrance
• Journey tier-perk hover flyout on Five Circles ladder
• Wait-game daily challenges reworked as overpowered power fantasies
• Swarm level-up fanfare + Super Saiyan resume burst; active passives shown on pause
• Swarm health pickups and XP gems never despawn; pixel-art boon pickups and HUD chips
• Swarm move speed buff; level-up freeze fixed; full-HP players skip medkit auto-pickup
• Jet Strike Powerstorm summons more enemies
• Deep Research article list cards styled for lite mode

2026-06-05

• Q&A popout drawer on extractions and Deep Research articles — mobile bottom sheet with swipe dismiss
• Synthesis Copy button visible on mobile
• /play settings menu closes when you tap outside

2026-06-04

• Swarm Orbit Shards reworked into spinning radial blades
• New Swarm passives (Slipstream, Precision) and boons (Bullet Time, Storm, Siphon, Vacuum, Nova, Overdrive)
• Swarm achievements and medals with pixel-art icons; challenge runs can grant power-ups
• Mine Drop detonates instantly when a boss is in blast range

2026-06-03

• 5 new Swarm enemies — Weaver, Mortar, Stinger, Burrower, Splitter
• Q&A thinking block renders properly in Quick Chat; cleaner list styling

2026-06-02

• Swarm Beam Lance auto-aims nearest enemy (not facing direction)
• /play — back button hardened; background no longer sends you to Journey; achievement popups kept
• Background music on /play and wait games
• Swarm — Brood, Bulwark, and Stalker elites; Toxic Monarch boss; balance tweaks
• Extraction synthesis reads more clearly (reader-first prompts)
• Notion export for any signed-in member

2026-06-01

• Play link in the main nav
• Improved game audio (Jet Strike reactive soundtrack, richer 2048 merge sounds)
• /play rotating backdrops; game cards above daily challenges
• Swarm — new survivor-style game on /play and during extraction waits (weapons, level-ups, bosses, challenges, pixel art, optional sound)
• Mind maps on extractions — generate from synthesis, saved on the extraction
• Paste a page URL to extract an embedded YouTube video; badge when the source was embedded; auto-strip trailing text after links; clearer errors for shortener or redirect links

2026-05-31

• Lite mode for slower devices — lighter visuals, with an optional manual toggle in settings
• Homepage feature deck expanded to 16 cards and translated; “Research” card renamed Academic Review
• Header nav centered; switches to icons-only on narrower screens

2026-05-30

• Score Attack — timed 90-second challenge runs on /play
• More daily/weekly challenge modifiers for Jet Strike and Space Invaders
• /play arcade fully translated
• +1 Wisdom Point when you accept a daily or weekly challenge
• Deep Synthesis restyled with updated gold branding
• Homepage loads faster on first paint
• Daily and weekly challenges on /play with leaderboards and cosmetic medals

2026-05-29

• New /play arcade and extraction wait games — play 2048, Jet Strike, or Space Invaders to earn Wisdom Points (15/day free-play cap); optional games while extractions run
• Space Invaders with boss waves, powerups, combos, keyboard or mouse control, and optional sound
• Jet Strike vertical shmup with powerups, combos, graze scoring, and optional sound
• Pause, resume, and fullscreen on /play; Play-games entry points on Journey and Profile

2026-05-28

• Spotify podcast extraction improved — smarter episode matching (Unicode titles, date/duration fingerprint, Podcast Index), verified YouTube fallback when no public audio feed, clearer “episode not found” guidance with a prefilled YouTube search link
• Extraction won’t start until Spotify finds a match for the episode link you pasted
• Spotify discovery warnings and errors translated across all languages

2026-05-27

• Fixed /discover content running too wide; fixed subtitle text clipped at the bottom; tips row fully translated
• Chat research accepts queries up to 16,000 characters
• Fixed research errors not displaying correctly in /chat
• Chat warns when you paste more than three links — toast plus inline reminder until you trim URLs
• TikTok, Instagram, and Facebook links in /chat and Deep Research now fetch transcripts when pasted
• Fixed members-only and restricted YouTube videos sometimes failing with a generic error after a long wait

2026-05-26

• /chat sidebar “Last 30 days” and “Older” groups start collapsed
• Mobile /chat: sidebar starts collapsed; guests see sign-in on Settings
• /chat shows when your weekly limit resets; new chats use your default persona
• Council session list dots now use each provider’s brand color

2026-05-25

• Homepage now has an interactive 3D feature deck — browse 10 product features (replaces the old value-prop grid)
• Council refresh — voice input in the composer, resizable sidebar (like /chat), collapsible model bars on the Models tab, session rename, and Cmd+\ to toggle the sidebar
• Chat settings now show weekly usage — combined token meter across all models
• Slides button on community share pages
• Community category pills start on a different random slice each visit
• Upgraded deep-mode Deep Research — new research engine with live “sources explored” and a scrollable sources panel on the article
• You can still open past Deep Research and Outside Perspective after your subscription lapses
• Documentation updates: Chat

2026-05-24

• Outside Perspective stays open after you generate it — reloads show your saved result; bookmarks jump straight to the section
• Homepage loads faster
• Chat sidebar resize handle is easier to find and works in dark mode
• Refreshed Chat documentation (skills, personas, links, research)
• Added voice input to /chat — dictate messages with a mic button in the composer
• /chat now analyzes links
• Added CSV and spreadsheet (.xlsx) upload to /chat and Council
• Uploaded documents are now preserved across /chat retry and edit
• Chat delete confirmation now dismisses when you click outside it
• Fixed the sticky nav header vanishing when nav dropdowns open mid-scroll

2026-05-23

• Added chat skills — a library of built-in and custom skills you can attach to a chat, with a skill picker

2026-05-22

• Added file and image upload to chat (paperclip, drag-and-drop, paste)
• Added research as a first-class chat turn with selectable depth and citations
• Added chat personas with a picker, editor, and prompt optimizer

2026-05-21

• Launched /chat — a single-model conversational surface with a full-catalog model picker, web search, and conversation archive
• Added full-text search and multi-select bulk actions to the chat sidebar

2026-05-19

• Fixed Notion export 400 error when an extraction had an empty videoUrl

2026-05-17

• Added a Snapshot callout at the top of deep research articles
• Q&A now redirects knowledge-gap answers to the globe web-search button
• Deep research articles now auto-fail after 10 min instead of hanging in polling
• Fixed citation-index misalignment in deep-mode research articles
• Reduced upload file size cap from 25MB to 7MB
• Shortened the upload zone header to just the verb across all locales

2026-05-14

• Models in the picker auto-deselect while their provider is reporting issues

2026-05-13

• Homepage loading speed optimizations
• TikTok/Instagram/Reddit search-upstream outages now surface clearer reasons instead of looking broken

2026-05-11

• Quick Access now surfaces research-article infographic, audio, and podcast content
• Added night-light theme support to /docs
• Browsers now force a cache-busting reload when they hold stale HTML after a site update
• Fixed Android PWA share target rejecting visits where the User-Agent contained "Mozilla"
• Fixed slide export errors from missing AI fields and NUL bytes in research articles

2026-05-10

• Sticky shortcut bar now hides once you reach the Scroll-to-top button
• Added a desktop split view for synthesis and Q&A — sticky side-by-side columns on extractions, deep research articles, and deep syntheses
• All users gained a free-week of CORE+
• Added action buttons (markdown export, infographic) to Deep Synthesis pages
• Quick Access: cached slide decks open directly without the style picker, and bookmarks refetch every 10s while the panel is open
• Slides loading dialog is now closeable with a background-hint indicator while generation continues
• TikTok URLs now have share-link tracking params stripped before extraction

2026-05-08

• Mobile Q&A: dedicated top bar with a prominent sidebar toggle
• "Copy All Responses" button moved into the Q&A header
• Fixed profile dropdown closing instantly after opening
• Added expanded Q&A mode with a cross-source history sidebar — full-screen Q&A with searchable history, archive support, and a mobile drawer
• Q&A web search now recovers from mid-stream Gemini rate limits
• Fixed Q&A web search occasionally failing with "No final result received from server"

2026-05-07

• Added a mobile share-sheet CTA to the homepage for iOS, Android, and desktop visitors
• Added a new /mobile install page with Android PWA Share Target and an iCloud Shortcut for iOS
• Linked the mobile install flow from the /extension page
• PDF/TXT upload toggle now appears on mobile
• Made infographic slides edge-to-edge
• GPT image slides now bookend the deck instead of clustering before the closing slide
• More grid items extracted into slides via prompt floors and a server-side fill safety net
• Fixed image caption cutoff and icon-card alignment in slides
• Fixed top/bottom clipping on infographic slides
• Fixed accent-as-text contrast on light slide presets (Vintage Editorial, Split Pastel)
• Audio cache now self-heals after site update
• Fixed stuck play button after track swaps

2026-05-06

• Added a floating "Save all" bar in Profile for cosmetic preferences
• Fixed Profile voice hydration bug and tab styling
• Fixed sparse final slides in Pro-mode decks
• Adaptive grid layout for icon-grid and annotation-diagram slides
• Fixed slide 0 reveal blur in iframe previews

2026-05-05

• Fixed Vintage Editorial slide title overflow and iframe preview blur
• Pro-mode slide imagery now driven by curated brief slices for better conceptual variety

2026-05-04

• Added Pro-mode slide decks to extractions — downloadable presentations with style picker, regenerate, and variant caching
• Added GPT Image 2 infographics to Pro-mode slide decks
• Slides design upgrade — spring physics, SVG line-drawing animations, refined AI prompts
• Co-host podcast now available on deep research articles — same dialog and global audio player as extractions
• Slides Regenerate button reopens the style picker instead of silently re-running
• Pro mode toggle redesign — custom checkmark, better visibility, softer borders
• Action button labels are now bold across extractions and deep research
• Extraction slides appear in Quick Access toolbar and bookmarks
• Fixed white-on-white variant dropdown options in slides dialog

2026-05-03

• Improved and streamlined audio/podcast process
• Highlights now use the Verdana font for better readability

2026-05-02

• Fixed legitimate public YouTube videos being blocked when YouTube bot-walled our scrapers
• Better detection of auth-required URLs (private AI chats like Claude/ChatGPT/Gemini/Poe, email, workspace tools, streaming services)
• Fixed the Begin button mobile layout on /deep
• Fixed a server crash when anonymous extraction errors hit the catch block
• Added a global Spotify-style audio player bar
• Audio and podcast bookmarks now auto-play when opened from Quick Access (also from the Feedback widget)
• The audio player's title and thumbnail now link back to the source extraction
• Linked Profile → Voice Preferences from the podcast dialog

2026-05-01

• Allow deleting individual audio variants from the Audio Library
• Better error messaging and a longer 200s timeout for Facebook, Instagram, and TikTok extractions
• PDF extraction now falls back more appropriately
• Added Grok 4.3 to the extraction model options
• Added GPT-5.4 and GPT-5.4 Mini to the OpenAI extraction picker, replacing GPT-5.2

2026-04-30

• Q&A answers can now include real [M:SS] timestamps quoted from the video
• Added visual analysis to Q&A on YouTube extractions
• Restyled Unique Model Perspectives as a clean table with model-colored row accents
• Replaced the plain # column in synthesis tables with a cosmic numeral badge
• Auto-recover from blank-page state with a service-worker reset and a visible recovery banner
• Inaccessible YouTube videos (private, members-only, region-blocked, removed) now fail with a specific reason instead of timing out
• Clearer error when a PDF URL serves an HTML page (login wall or viewer SPA)
• Q&A now replies in the same language as the question
• Fixed stuck SPA navigation between pages
• Fixed extraction summaries showing duplicated subsections

2026-04-29

• Added Highlights at the top of every extraction
• A wave of new achievements: Notion Connected, Notion Native, Full Treatment, Anniversary, Council Veteran, Citation Heavy, Weekend Warrior, Streak Saver, Research Scholar, Full House, Tastemaker, Active Reader
• Lowered Power User threshold from 150 to 125 extractions, and Keep Cooking from 15 to 14 days
• Fixed extraction summaries where section headers fused into the body text
• Fixed the # column floating wide on Common Themes tables that lacked a Title/Description split

2026-04-28

• Common Themes table now stacks vertically on mobile, and the redundant column headers were removed
• Restyled All Purpose synthesis themes and takeaways as clean, scannable stacked tables
• Journey page now shows your tier, total wisdom points, and unlocked achievements next to branded section headings on desktop
• Unreachable URLs now fail in under a second instead of hanging for a 30-second timeout

2026-04-27

• "Compare Reports" button label shortens to "Compare" on mobile, and the top action buttons are larger on desktop
• Notion export now works for fast and deep research articles
• Fixed silent extraction failures caused by some AI providers returning unusual response shapes
• Fixed extraction sometimes getting stuck for caption-less YouTube videos
• Podcasts now have a two-voice co-host option — pick two AI voices for a NotebookLM-style dialogue
• Podcasts now appear in Quick Access bookmarks alongside audio

2026-04-26

• Launched the CoreWise documentation site at /docs
• Added a "What can CoreWise do?" discovery banner and Docs footer link
• Fixed PDF extraction failing for GitHub file URLs
• Web search in Q&A is now live for CORE+ and The Core tier members
• Q&A attachments now accept Word documents (.doc,.docx), plain text, and markdown files
• Q&A suggested questions panel now has regenerate (up to 3 times) and hide controls
• Pasting a YouTube search results URL now shows an inline warning with a link to Discover instead of failing

2026-04-25

• The upload button now accepts Word documents (.doc,.docx), Markdown, and text files in addition to PDFs

2026-04-24

• Upgraded OpenAI models to GPT-5.5, replacing GPT-5.4 Mini as the default
• Added GPT-5.5 (xhigh) as a new extraction option for maximum reasoning depth
• Fixed follow-up question suggestions silently failing on extraction and deep research pages
• Fixed Council Chat synthesis disappearing after sending a follow-up question

2026-04-23

• Added rotating category filters to the Community page, refreshed weekly

2026-04-20

• Fixed markdown export showing `[object Object]` in the Unique Model Perspectives section
• Fixed audio export going silent on research paper syntheses
• Added a Connect Notion shortcut directly on the Profile page
• Notion toolbar button now navigates to setup when Notion isn't connected yet
• Mobile layout improvements: channel grid and share bar no longer overflow

2026-04-17

• Fixed legitimate health, scientific, and medical content being incorrectly refused

2026-04-16

• Community-published extraction pages now have share buttons (Twitter, LinkedIn, Facebook, WhatsApp, Email)

2026-04-12

• Fixed YouTube links copied from Google search results failing to extract

2026-04-07

• Fixed streak counter not updating until your next extraction, and streak protectors now cover exactly one missed day

2026-03-28

• Old or cached model selections are now auto-corrected instead of causing extraction failures

2026-03-27

• Infographic generation now retries automatically on temporary failures

2026-03-26

• Private and restricted Instagram content now fails quickly with a clear message instead of timing out
• Fixed a crash when opening older extractions without Unique Insights

2026-03-25

• Replaced Llama 4 with the new Mercury 2 from Inception

2026-03-23

• Fixed visual glitches in 3D effects when the WebGL context wasn't ready
• Fixed comma-separated citation references (e.g. [25, 26]) not rendering as clickable links

2026-03-22

• Fixed server crashes after extended uptime caused by memory growing without limit
• PDF extraction now falls back to Gemini when PyMuPDF and Mistral OCR both fail

2026-03-19

• Fixed Instagram reels longer than 2 minutes failing to extract
• Added Grok 4.20 Reasoning as an extraction model option
• Improved Core Message quality — extractions now lead with concrete details
• Fixed Danger Zone text readability in night-light theme

2026-03-18

• Fixed extraction errors with GPT-5.4 Mini

2026-03-17

• Upgraded models: GPT-5.4 Mini (was GPT-5 Mini) and Mistral Small 4 (was Small 3.2)
• Auto-corrects malformed URLs with a duplicated prefix when pasted

2026-03-16

• YouTube multi-video collection URLs now show a warning instead of failing

2026-03-13

• PDF extraction now falls back to Mistral OCR when the PDF service crashes

2026-03-12

• Deep research now searches Tavily in parallel and shows those sources in the Sources section
• Fixed sign-in page styling in night-light theme

2026-03-11

• Fixed missing citations in Perplexity deep research results
• Fixed extraction failing on image-only Instagram posts

2026-03-10

• Fixed voice input not working in Firefox
• Newest released models in Council, including Grok 4.20 Beta
• Improved slide generation reliability with retry logic

2026-03-09

• Added file upload (images + PDFs) to Q&A chat and Council Chat
• Added manual "Run Synthesis" button to Council Chat
• Added shimmer animation to the Debate and Run Synthesis buttons

2026-03-08

• Fixed article and deep research citations rendering as raw URLs instead of [N] markers
• Added Profile and CORE+ links to the mobile nav
• Improved bookmark drawer section heading visibility

2026-03-07

• Fixed self-serve account deletion
• Added AI model overload alert to the API status banner
• Fixed multiple night-light theme visual issues — animated rings background no longer bleeds through on mobile or light-theme pages
• Disabled 3D background sphere on mobile screens
• Fixed citation markers in articles displaying as raw URLs instead of clean footnote references
• Fixed slides preview showing a cut-off header and intermittently blank content
• Suggested Q&A questions now use simpler, more accessible language
• Removed fullscreen button from the slides toolbar
• Improved readability of the Danger Zone section in Settings
• Fixed extraction slide text being cut off
• Slide content area now smoothly animates its height to fit incoming text, no more jarring layout shifts

2026-03-06

• Upgraded extraction engines to GPT-5.4 across all AI pipelines
• Added Inception Mercury 2 to Model Council defaults
• Restored Q&A Explore Topics suggestions — section was silently returning no results due to an invalid model ID, now fixed with GPT-5.4
• Upgraded council synthesis to OpenAI Responses API with reasoning
• Added Tier Showcase to the Journey page above the "How to Earn Points" section
• Journey page desktop headings are larger; "Unlock more" now links directly to the CORE+ page
• Improved community page scrolling performance — reduced Framer Motion overhead for off-screen cards
• Improved server startup reliability — health checks no longer return 500 or connection refused during initialisation

2026-03-05

• Added unread support message badge on user nav button
• Added unread count in Manage CORE+ dropdown link
• Enabled deep thinking (HIGH) for all Q&A — Flash Lite now uses extended reasoning
• Fixed community page performance degradation over time
• Fixed avatar animation overload and thumbnail 404 cascades on community feed
• Replaced static avatars with CSS containment for community feed
• Fixed TLS socket errors from Neon PostgreSQL idle connection termination
• Migrated Gemini SDK to new unified Google GenAI SDK
• Fixed ghost blur on slide transitions
• Locked Recap slide content to fixed height to prevent layout shift
• Updated Cited Intelligence card description for Fast mode
• Updated Deep Search card description on deep research page
• Consolidated metadata cache sweep into daily cleanup job

2026-03-04

• Replaced Grok 4.1 synthesis engine with GPT 5.2 (thinking)
• Added "Specific Details" section to synthesis — includes a knowledge-chain timeline
• Added markdown table rendering to PDF exports
• Added night-light theme support for PDF exports
• Notifies users when Claude is overloaded and synthesis falls back to a backup AI
• Styled share controls: larger icons, Community label, matches Copy Link to Notes style

2026-03-03

• Added Perplexity deep research content navigation sidebar on research articles when Perplexity content is being shown
• Clickable [N] citation markers in deep articles — styled as small superscripts, adjacent markers fixed
• Content-aware suggested questions in Q&A panels (deep research + fast research), generated in the user's selected language
• Per-model CORE+ requirement toggle in Model Controls panel
• Switched synthesis engine from Grok 4.1 to GPT-5.2 (Thinking)
• Fixed deep articles not auto-translating on load when a language is selected
• Fixed Google redirect URLs embedded in third-party site paths

2026-03-02

• Extraction title can now be edited — inline pencil icon, saves on blur, always visible on mobile
• Web article title fix — replaced with parallel webpage title fetch for accurate titles
• Built AI-generated image slides for Nana Banana 2 mode using Gemini image model (preview — UI coming soon)
• Updated tier comparison table with Outside Perspective, Mind Maps, Slides, Bulk Processing, and Extended Length Limit
• Renamed tier showcase labels for clarity: Full Context Q&A, Perplexity Deep Research, Extended Length Limit
• Reduced Seeker council debate limit from 5 to 3 per month
• Restricted Mind Map and Slides features on deep research articles to logged-in users
• Added help bot loading indicator for Slides, Infographic, Audio, and Podcast generation
• Made synthesis model count messages dynamic — shows actual successful model count
• Added Slides generating translation to all 19 non-English languages

2026-03-01

• Launched interactive Tier Showcase on CORE+ page — clickable tier orbs with custom SVG icons, spring physics, and full feature comparison
• Added Deep Synthesis, Extract Later, and latest Council session shortcuts to Quick Access toolbar
• Implemented monthly usage quotas for Slides, Mind Maps, Audio generation, and Podcast exports
• Added mode-aware quota card on Deep Research page — surfaces upgrade prompt when monthly limit is reached
• Updated Terms of Service with content safety guidelines, accuracy improvements, and criminal content prohibition
• Reduced extraction limits: Explorer 15→10/month; Curator and Sage limits adjusted across codebase
• Reduced deep research sub-limits for Explorer, Curator, and Sage tiers
• Added achievement links to Journey page gems
• Hidden Library/Discover nav labels at tablet widths to reduce clutter
• Lazy-loaded TierShowcase, galaxy SVG, comparison table, and FAQ on CORE+ page for performance
• Fixed render crash on /coreplus page (hooks-in-IIFE violation)
• Fixed race condition in council guest quota tracking
• Filtered API provider status warnings to only show on relevant pages

2026-02-28

• Added LinkedIn post and article extraction support
• Overhauled infographic pipeline: Flash 3 content curator, prompt quality rewrite, and Google Search grounding
• Converted infographic images from PNG to WebP for smaller file sizes
• Blocked shortened TikTok URLs (vm.tiktok.com) with inline validation

2026-02-27

• Redesigned Community page with matched Library width, 3-col grid, and new filter controls
• Enhanced synthesis prompts with named individuals, organizations, and precision rules
• Fixed bulk extraction failing for non-YouTube videos (Instagram, Facebook, TikTok)

2026-02-26

• Upgraded infographic model to gemini-3.1-flash-image-preview
• Made DeepL primary for synthesis translation with bold and newline preservation fixes
• Restricted deep research mode to registered users — guests can only use Fast
• Renamed Grok labels to 'Grok 4.1' + 'Fast Reasoning' across profile and homepage
• Added scroll-to-top button above Q&A section on extraction detail page
• Added i18n translation support for synthesis action buttons
• Fixed deep research citations not merging search_results URLs with inline citations
• Fixed audio export for translated synthesis (markdown + section labels)

2026-02-25

• Added Quick Access bookmarks panel with right drawer, help bot integration, and recent extractions
• Added infographic generation to deep research articles with button and i18n support
• Added elegant sources popover to Deep Research library list
• Replaced Mistral Large 3 with Grok 4.1 Fast Reasoning as synthesis engine
• Added Compare button to lower inline button area for extraction reports
• Added Profile link to mobile/tablet menu
• Added signup-prompt button bar to guest ExtractionPreview
• Added cosmic space vibe for Quick Access button in feedback widget
• Wrote deep articles in the query's language via Layer 4 directive
• Persisted extraction state across page refreshes
• Pre-warmed bookmarks/extractions queries on widget open for faster access
• Updated profile CORE+ description and highlight feature text across all languages
• Trimmed auth page copy and reverted headings to Chaney Ultra Extended
• Fixed sticky menu Audio, Podcast, and Infographic button access and quota gating
• Fixed infographic button in deep library not responding to clicks
• Fixed double-retries on 503 and improved overloaded backoff
• Fixed auth page layout shift when stats block loads in
• Fixed sticky mini toggle visibility until user scrolls past primary CTA
• Closed feedback widget on click-outside
• Added Infographics to /coreplus Creative Tools feature list
• Kept Compare Reports panel collapsed on article load/refresh
• Fixed auth page mobile clipping of stats and footer

2026-02-24

• Added AI infographic generation for extractions using Nano Banana Pro
• Added API provider status indicator banner
• Added sticky CORE+ orb CTA and replaced secondary upgrade button
• Added 4 new feature-selling signup options to all 20 i18n files
• Added Load More pagination to deep research library to fix missing articles
• Made infographic limits tier-based (3/5/7/10/30) and refined dialog UX
• Enhanced slide presets with modern design aesthetics and enabled Slides button
• Renamed divergence to unique perspectives with tension level classification
• Updated deep article revision prompts with Outside Perspective framing
• Fixed PDF extraction for wrapper/viewer URLs with embedded PDF query params

2026-02-23

• Added Comparison button for deep research (Sonar Deep Research article vs Final Synthesized Article)
• Redesigned slides: kinetic storytelling, art gallery picker, screening room viewer
• Added i18n support for achievement names across all 20 languages
• Converted top action bar to 4-per-row grid with larger buttons on desktop
• Made deep article inline buttons larger with 4-per-row grid layout
• Optimized /api/community-stats: combined 3 queries into 1 with 5-min cache
• Navigated to /coreplus when clicking locked CORE+ models in council
• Fixed NO_CAPTIONS error for videos with spoken words but no caption tracks
• Fixed transcript concatenation when transcript_only_text is missing
• Fixed SlidesStyleDialog invisible overlay issue
• Fixed chunk load error detection for 'Unexpected end of input'

2026-02-22

• Added 6 new starter achievements to promote feature discovery with custom SVG icons
• Upgraded CorePlus page to spring-physics, cursor-reactive design system
• Added underwater Three.js background with orbital camera rotation to Discover page
• Added animated galaxy SVG overlays to Journey and CorePlus pages
• Added Nebula Badge with float-in animation to CorePlus page
• Added generative wave art animation to homepage
• Added animated sun/moon toggle CTA button with cosmic theme to CorePlus page
• Added tunnel shader hover effect to CorePlus feature CTA cards
• Replaced static CorePlus nebula SVG with animated version
• Replaced Discover page subtitle and tips with styled SVG
• Disabled URL bar with scanning animation during extraction preflight
• Doubled deep council button sizes and fixed send button z-index
• Fixed OOM crashes by lazy-loading PDF service and capping memory
• Improved captionless video handling by skipping Supadata when captions are unavailable

2026-02-21

• Added 6 custom SVG icons for About page feature cards with doubled icon sizes
• Updated Deep Research mode card descriptions and copy for clarity
• Updated help bot with latest features and conciseness directive
• Moved quota usage text to the right of Library heading
• Fixed layout shift when switching between fast and deep mode cards
• Fixed convergence/divergence text cutoff on About page mobile
• Increased About page stat and START button SVG sizes on mobile
• Made Forgot Password link bold and larger
• Fixed ScapeCreators 400 Bad Request by removing invalid sortBy option

2026-02-20

• Added Research Slides — generate and present HTML slideshows from deep research articles
• Added animated feature cards replacing flat tab descriptions in Deep Research
• Added embedded Bottom Line mini-slideshow to extraction results
• Added rich animations and ambient visuals to slide presentations
• Used Gemini Flash 3 for server-side Bottom Line slides generation
• Wired up session sidebar in Council chat on Deep Research page
• Added Split View button to Models tab (not just Synthesis)
• Replaced debate model scorecard with research-focused insights scorecard
• Enforced scannable, no-wall-of-text formatting in debate verdict
• Doubled Synthesize Debate Insights button size
• Added i18n for remaining hardcoded strings in Deep Research and FloatingActionButton
• Doubled Audio Library header and toggle text sizes
• Widened Library page and switched to 4-column grid on desktop
• Optimized extraction loading scene for mobile performance
• Simplified extraction long-wait warning (removed phase/progress display)
• Updated Deep Research feature card descriptions for accuracy
• Fixed Core Synthesis appearing as flat paragraph instead of structured sub-sections
• Fixed slides viewer blank iframe (srcdoc instead of blob URL)
• Fixed Gemini malformed JSON error in slides generation

2026-02-19

• Added Model Council Debate Arena — models argue and refine perspectives on your topic
• Redesigned Council desktop layout with tab system, sessions sidebar, and full-width views
• Overhauled translation pipeline — Gemini primary with DeepL fallback for extraction translations
• Added pre-query Web Search and Auto-Debate toggle buttons to Council
• Added post-debate CTA to transition into AI synthesis with debate context
• Added rich typography and table styling for debate content
• Added synthesis section header translations with bold formatting preservation
• Enhanced model panel headers with provider logos, colors, and visual depth
• Scaled up debate UI on desktop: bigger headers, icons, and text
• Increased Council font sizes ~20% on iPad+ for better readability
• Added scrolling during streaming with animated scroll-down pill
• Added draggable resize handle back to split view
• Added custom subtitle for Council mode on Deep Research page
• Increased 'Tap to Ask' suggestion text sizes on desktop
• Reduced auto-debate to 1 round (manual debate remains 2 rounds)
• Upgraded Gemini Pro model from 3 to 3.1 in extraction dropdown
• Hidden Gemini 3 Pro Preview from council (superseded by 3.1 Pro)
• Added Deep Research CTA translations for all 19 non-English languages
• Improved debate critique prompts to reduce hallucinated criticisms
• Fixed council follow-ups referencing invisible search results
• Fixed scroll lock during streaming with wheel/touch detection
• Fixed synthesis tab disappearing on conversation reload
• Fixed synthesis vanishing on tab switch
• Fixed split-view unmounting synthesis panel

2026-02-18

• Added full-screen immersive council session view
• Added Copy All button to Source Analysis section
• Added web search toggle to Model Council
• Added Optimize Prompt button in Council mode
• Added guest limits for Model Council (1 session, 5 inputs)
• Revised Gemini Flash 3 synthesis for improved quality
• Fixed extraction results recovery when server completes during poll timeout

2026-02-17

• Launched Model Council with two-panel layout, AI synthesis engine, and multi-model streaming
• Added deep research articles to Community page
• Added AI-powered synthesis prompt suggestions to Council
• Added 'Save as default' for Council model selection
• Unlocked Council for all users with CORE+ badges for premium models
• Added mobile-responsive council layout with accordion and collapsible models
• Lowered Notion export requirement from Curator to Explorer tier
• Added i18n translation support for deep article view buttons and labels
• Replaced Claude Sonnet 4.5 to 4.6
• Fixed PDF service recovery with proactive restart on connection errors
• Fixed Helvetica-Bold italic font resolution error in PDF export

2026-02-16

• Added article notes feature with auto-saving drawer, send-to-notes from highlights, and markdown preview
• Added notes feature to extraction detail pages with send-to-notes from highlights
• Added server-side PDF export for Deep Research articles
• Added sticky shortcut menu for extraction detail pages
• Added voice input (speech-to-text) to all Q&A chat bars with Vosk fallback for Firefox
• Added Creative domain focus for artistic and creative content synthesis
• Added YouTube video refresh button to deep research articles
• Enhanced deep article Q&A with full pipeline context and web search
• Added Export dropdown to /deep articles replacing standalone Markdown button
• Added Ask button to SynthesisView header and sticky menu
• Added mid-article Ask Anything button with shimmer effect
• Added Ask Anything icon to sticky menu in deep articles
• Made sources count clickable to scroll to Sources section
• Reorganized toolbar buttons into inline article section with responsive grid
• Added Notes and Project shortcut icons to sticky audio/podcast bar
• Added CSS flickering grid background to deep research loading box
• Moved Mind Map icon next to sources in library cards
• Added i18n support for AchievementToast in galaxy loading sequence
• Replaced browser confirm() with styled popup for highlight management
• Made Q&A response boxes full width on desktop and mobile
• Fixed CorePlus heading auto-scaling for translated pages
• Improved Facebook video extraction with quality checks and transcript caching
• Fixed Q&A streaming crash with Google Search grounding enabled

2026-02-15

• Added voice input (microphone) for Deep Research queries
• Added text highlighting to deep research articles
• Added audio & podcast export to Deep Research articles with sticky floating bar on scroll
• Redesigned About page with hexagonal 6-model visual and updated convergence footer
• Added interactive particle vortex effect to CORE+ page
• Added neon flow hover effect to /deep research textarea
• Made deep research articles full-width on mobile
• Show audio/podcast buttons to all users with tooltip + redirect to CORE+
• Added i18n translation support to CORE+ Manage page across all 20 languages
• Improved wisdom extraction prompts: removed conciseness constraints, added structured tables
• Replaced Three.js WebGL effects with lightweight CSS alternatives
• Added non-streaming fallback for Gemini stream parse failures

2026-02-14

• Added Audio Library section to Library page
• Added Model Council: multi-model streaming chat (coming soon)
• Added full-screen guest cooldown CTA overlay on homepage with benefits messaging
• Added rotating tips section below Deep Research search bar with i18n support
• Added rotating gold border effect to Discover, Research, and Deep Research search bars
• Added Deep Research to mobile menu
• Added 'Change voice' link in Audio Library to profile voice settings
• Added night-light theme overrides for Deep Synthesis and Audio Library
• Caught TikTok regional homepage URLs as unsupported
• Fixed deep-research timeout causing results to be wasted
• Fixed PDF extraction crash: handle pymupdf4llm internal errors gracefully

2026-02-13

• Added fallback for Instagram transcript extraction
• Added inline URL warning in Deep Research queries with i18n support
• Added Mistral as intermediate synthesis fallback between Claude and Gemini
• Added search filter translations across all 19 non-English languages
• Updated direct media extraction descriptions to include Spotify podcasts across all languages
• Updated deep research limits with international translations
• Made tooltip info icons more visible with larger size and amber color
• Hardened client against Replit proxy transient errors (504, 429, non-JSON)
• Fixed Gemini 429 rate limit failures with longer backoff and Flash→Pro fallback
• Fixed YouTube channel tab URLs not caught by unsupported pattern
• Fixed Instagram extraction NETWORK_ERROR failures
• Fixed deep research articles showing gradients instead of thumbnails

2026-02-12

• Added X (Twitter) platform support for tweet extraction via ScapeCreators API
• Added Perplexity search filters (domain and time) to Deep Research UI
• Auto-resume deep research polling after tab close or page refresh
• Added Mistral OCR fallback for image-based/scanned PDFs
• Added translation support for deep research articles (/deep/:id)
• Modernized /deep page dropdowns with Radix UI Select and tooltips
• Reorganized /deep homepage: Projects above Library, added source count and Mind Map button
• Added guest-specific signup CTA to usage limit screen
• Improved /deep UI with animated buttons and post-optimize state
• Added translations for Ask Anything and Deep Research on CorePlus page
• Updated fast research timing estimate from ~60s to ~90s across all 20 languages
• Enabled prompt caching for all Claude extractions for faster responses
• Deep Research Fast mode: replaced L2 Mistral Small with Mistral Agent (Magistral Small + Premium Search)
• Optimized model reasoning effort for faster deep research responses
• Blocked Vimeo and Google URLs with inline warnings (unsupported platforms)
• Fixed Mistral Conversations API streaming format parsing
• Fixed dropdown readability on deep research filters

2026-02-11

• Implemented overlapped deep research pipeline - ~30-60s faster deep mode extractions with full model streaming
• Added 6 new achievements: Highlighter, Assembly Line, Rosetta Stone, plus starter/advanced for Extract Later, Research, and Audio
• Enabled web search grounding for GPT-5.2, Grok 4.1, GPT-5-mini, and Gemini Flash 3 across deep mode layers
• Made Fast Research the default mode with bold time estimates
• Translated Discover and Research nav items across all 20 languages
• Added inline URL validation for TikTok profiles, Instagram image posts, and Bing video URLs
• Comprehensively fixed night-light theme readability on deep research articles
• Increased content map (mindmap) font size to 17px for better readability
• Fixed duplicate perplexityData declaration crashing app on load
• Improved handling of Anthropic 529 overloaded errors
• Added graceful fallback for async WebGL context creation failures

2026-02-10

• Added TED talk transcript extraction with YouTube fallback
• Added 3 new starter achievements: Research Pioneer, Pathfinder, Personal Touch
• Reworked community Hot filter to prioritize views over upvotes
• Updated /deep tagline to mention Mindmaps across all 20 languages
• Prevented browser extension removeChild/insertBefore errors from crashing app
• Improved the fallback for captionless YouTube videos - videos without built-in captions can now be transcribed and extracted
• Added share links bar to deep research article pages
• Upgraded deep research article Q&A ('Ask Anything') to match extraction Q&A experience
• Added CORE+ upgrade CTAs to deep research for quota limits and tier gating
• Added Ask Anything to CoreWise Essentials on CORE+ page
• Added Perplexity deep-research polling timeout (2.5 min) with sonar-pro fallback for improved reliability
• Added BETA tag to Deep Research heading
• Moved Deep Research CTA from Discover to Research page with matching style
• Added Mistral-powered YouTube query optimizer for improved search results
• Added aggregation fallback to filter failed model outputs and improve synthesis reliability
• Optimized model selection: updated fast mode with fresh models and refined deep mode search
• Fail fast on empty transcripts during preflight instead of after galaxy loading for better UX
• Fixed false positive empty transcript detection by trying Supadata fallback first
• Fixed multiple crashes: sessionStorage in restricted browsers, Radix Select + AnimatePresence race conditions, and route transitions
• Fixed Safari compatibility by replacing regex lookbehind with capture group
• Fixed guest header on /deep always visible for non-authenticated users
• Fixed model dropdown rendering in two places on homepage
• Fixed back navigation failing after first deep research article load
• Added UUID param validation to prevent 500 errors from bot/crawler requests
• Fixed CSS preload failure detection in lazy route loading
• Improved deep research article responsiveness and mobile UX

2026-02-09

• Launched /deep research page system - generate comprehensive standalone research articles on any topic with shareable permalinks
• Added AI-generated interactive mind map visualization for Deep Research articles
• Added Content Navigation feature with full-width mindmap and scroll-to-section functionality
• Enhanced deep research article typography with gold accents and styled JSON code blocks
• Added mid-article Mind Map CTA and Scroll to Top button for easier navigation
• Added full i18n translation support for Deep Research pages across all 20 languages
• Converted LaTeX math notation to Unicode for better readability
• Added default OG image for social media sharing
• Updated homepage copy: four perspectives → six perspectives
• Simplified nav bar with icon-only '+ New' button

2026-02-08

• Added Mistral Large 3 as synthesis engine option
• Enhanced Q&A feature to use Perplexity deep research context for richer, more comprehensive answers
• Added Llama, Grok, and Mistral orbs to galaxy extraction loading scene with 6-model visualization
• Improved synthesis quality with updated system prompts and removed artificial sentence count limits
• Added 'Synthesized by' model indicator to Source Analysis section
• Fixed deep research not being passed to Q&A for completed extractions
• Fixed mobile layout issues with extraction title and orb positioning

2026-02-07

• Added Grok (xAI) and Mistral as new extraction model options with updated 2-row model selector UI
• Enabled Research Paper domain focus - now available for extracting academic papers with specialized prompts
• Added Perplexity Sonar Deep Research integration - comprehensive multi-source research with citations and clickable links
• Added Web search tab to Discover page using Perplexity API - search the web for articles and content
• Added auto-rotating topic suggestion chips to Discover page empty state with 20 topics per platform
• Added custom Deep Research icons and improved UI alignment across synthesis sections
• Added auto-reload on stale chunk errors after deployments for seamless updates
• Improved Web preview modal with full-screen layout and markdown rendering
• Always show filters on Discover page for easier access

2026-02-06

• Added Extract Later feature to Discover page - save content for future extraction with one click
• Improved Help Bot with expanded knowledge base, conversation memory, and user context awareness
• Enhanced Discover page UX with Quick Tips, search bar styling, and cleaner video cards
• Added top-level error boundary to prevent blank page errors and improve error logging

2026-02-05

• Added 14 new custom avatar designs including sacred geometry, torus, and creative variants with tier restrictions
• Added Gemini Flash 3 fallback system to automatically switch to 3 Pro when TPM limit is temporarily breached
• Added current voice selection display with supported languages on profile page
• Improved PDF upload reliability with timeout, retry logic, and automatic service recovery
• Added dynamic font sizing for long extraction titles to improve readability
• Fixed text selection being reset during Q&A animations with memoized markdown rendering
• Fixed preview hook extraction to handle GPT heading and plain text formats

2026-02-04

• Added toast notifications when Audio/Podcast generation buttons are clicked with full i18n support
• Fixed model names being stripped from unique insights text

2026-02-03

• Added Reddit subreddit search feature to Discover page - search within specific subreddits
• Updated YouTube search with popular sort option and improved filtering
• Increased extraction timeout to 7 minutes with long-wait warning for better reliability
• Improved error messages for geo-restricted and unavailable videos

2026-02-02

• Improved Q&A chat UX with copy-as-markdown buttons, Copy All Responses, and optimized window heights
• Enhanced Q&A mobile experience with full-width response bubbles
• Fixed OpenAlex search crash on queries with special characters
• Added retry logic for transient database connection timeouts

2026-02-01

• Added OpenAlex research paper discovery feature - search and extract academic papers directly
• Added full-page tablet menu for screens ≤1300px with improved navigation
• Added i18n translation support for Discover and Research pages across all 19 languages
• Added rotating topic suggestions and enhanced UX on Research page
• Added SafeLinks URL decoder to handle email security wrapper URLs
• Redesigned Research page to match Discover page design system
• Fixed missing share options (Copy Link, Community toggle) for PDF extractions
• Fixed DOI redirect triggering false login-page detection

2026-01-31

• Added Domain Focus feature with Research Paper and Tutorial modes - custom synthesis prompts optimized for academic papers and how-to content
• Added 90-day webpage content cache for faster article re-extraction
• Added comprehensive i18n translations for extraction loading, synthesis engine, domain focus, QA chat, and Deep Synthesis features
• Fixed multiple bold formatting issues in tutorial synthesis output

2026-01-29

• Improved mobile responsiveness: About page stats, Start Extracting button, and Take Guided Tour button

2026-01-28

• Added 9 new achievements: 3 Starter, 3 Advanced, and 3 Rare tiers
• Rebalanced achievement requirements for better progression (Community Favorite, Global Impact, Philosopher, Speed Demon)
• Added Spotify to platform logo rotation on homepage with demo URL
• Updated CORE+ page with improved feature descriptions and FAQ refinements

2026-01-27

• Added Spotify podcast support - automatically converts Spotify episode URLs to RSS feeds for extraction
• Added transcript spelling correction using video/podcast titles for improved accuracy
• Improved extraction reliability with retry logic for network failures and optimized timeouts
• Added monthly extraction limit reset date display on profile page
• Fixed Together/Llama source analysis visibility in Community view
• Added login page detection for webpages to prevent extraction failures

2026-01-26

• Added text chunking for ElevenLabs audio export with improved robustness
• Added livestream detection to reject livestreams early and prevent extraction timeouts
• Added PDF URL support for anonymous extractions
• Added inline warnings for sites with known scraping issues
• Fixed login crash when OAuth users attempt password login
• Fixed Deep Synthesis popup text colors on night-light theme
• Fixed bold formatting issues in Unique Model Perspectives

2026-01-25

• Bug fixed: PDF support on bulk extraction
• Added Instagram profile discovery to Discover page with URL validation and redirect support
• Increased transcript fetch timeouts for improved reliability
• Strip tracking parameters from article URLs for cleaner extraction

2026-01-24

• Updated Llama 4 model selection - removed Scout Instruct, kept Maverick Instruct

2026-01-23

• Lowered The Core tier wisdom requirement from 5000 to 2500 points
• Added Firefox addon support with official Firefox logo
• Fixed button text color visibility issues on night-light theme
• Fixed 413 entity too large error for extraction requests

2026-01-20

• Fixed database connection pool exhaustion causing server stability issues
• Enhanced Chrome extension YouTube button detection
• Prepared Firefox extension for Mozilla Add-ons marketplace with security and performance improvements

2026-01-19

• Added comprehensive multilingual support across homepage, Library, Community, Discover, About, and CORE+ pages
• Fixed markdown formatting issues with bold text rendering incorrectly in extractions
• Improved mobile UX with 'Start Extracting' link in profile menu and fixed rotating slogan text cutoff

2026-01-18

• Added multi-platform discovery system with YouTube, TikTok, Instagram, and Reddit search
• Comprehensively improved PDF processing with HTTP header-based detection, enhanced error diagnostics, and arXiv URL handling
• Improved Discover page UI with updated styling, platform brand icons, search placeholders, and 24-hour result caching
• Enhanced search bar UX with mobile-friendly layout and improved button positioning
• Comprehensive theme and visibility fixes across dark theme, night-light theme, and backdrop blur effects
• Updated preview modal to remove header text for all platforms
• Mobile navigation optimizations to save screen space
• Fixed YouTube metadata display in Discover page and eliminated jarring double-animations when switching platforms

2026-01-17

• Added translated audio support - generate synthesis audio in multiple languages
• Added Llama 4 Maverick Instruct model via Together.ai - available to all users
• Further optimized video extraction speed with improved transcript fetching and caching
• Added Gemini 3 Pro Preview model option
• Reordered GPT models to feature GPT-5.2 more prominently
• Updated feature descriptions for better clarity
• Fixed Llama model not appearing in Unique Model Perspectives

2026-01-16

• Significantly improved video extraction speed by optimizing preflight checks
• Added multilingual video transcription support for all languages
• Improved YouTube transcript handling for long videos with automatic fallback system
• Fixed critical bug with incorrect model attribution in Unique Model Perspectives
• Updated advanced model timing message to show ~1 minute instead of 'a minute or two'
• Improved extraction button layout and animations for better visibility

2026-01-15

• Added night-light theme support to Dev Log page for better readability
• Added embedded video to About page explaining CoreWise takeaways
• Fixed YouTube video embeds with privacy-enhanced player and proper styling
• Added mobile toggle for collections sidebar to improve small screen experience
• Updated homepage rotating slogans with clearer messaging and better timing
• Reorganized extraction page buttons into two rows for better layout
• Fixed critical server stability issues and improved error handling
• Optimized Community page hover performance to eliminate lag
• Added Clear Old Logs button to admin dashboard for easier maintenance
• Simplified page animations for better performance across all devices
• Fixed temporal dead zone errors on Library page

2026-01-14

• Optimized Journey page load performance with progressive lazy loading and staggered animations
• Comprehensively improved mobile responsiveness across navigation, buttons, and feature displays
• Added server-side error logging system for improved debugging and monitoring
• Improved application stability with optimized database connection settings
• Reduced extraction timeout from 15 minutes to 7 minutes for better resource management
• Added Google OAuth login for streamlined authentication
• Added ability to claim guest extractions after logging in with Google
• Improved homepage performance with lazy loading and deferred API calls
• Fixed layout shifts on homepage with improved skeleton loading
• Added smooth fade-in animations to homepage feature boxes
• Removed unused fonts to improve page load speed

2026-01-13

• Rebranded 'Full Steps' feature to 'Full Guide' for clarity
• Significantly optimized synthesis prompts for better AI output quality and readability
• Applied major extraction prompt optimizations to all AI models (Claude, GPT, Gemini)
• Improved bold formatting in AI-generated extractions for better readability
• Added Full Guide feature to platform tips blog post with examples
• Updated CORE+ feature descriptions for better clarity
• Added URL validation warnings for YouTube channels and platform homepages
• Updated extraction loading sequence text for better clarity
• Updated welcome email content and support contact information

2026-01-12

• Added 'Full Steps' extraction feature for tutorial and how-to content
• Reorganized extraction action buttons into two rows for cleaner layout
• Increased button and icon sizes for better visibility
• Added color coding to extraction buttons (podcast in orange accent)
• Optimized AI extraction prompts for improved performance and caching
• Added translation support for Journey page across all languages
• Updated tooltips for podcast and Full Steps features with clearer descriptions
• Changed 'Advanced Settings' to 'Customize' on homepage for simplicity
• Fixed tooltip text color for better readability
• Improved PDF export formatting for bullet points and headers
• Fixed galaxy background visibility on limit pages
• Added epistemic status markers to Outside Perspective feature
• Updated profile section with improved UI elements
• Removed redundant Share Link toggle for cleaner UX

2026-01-11

• Added shareable link feature for private extractions
• Auto-enabled link sharing for all social share buttons
• Disabled username onboarding popup on mobile screens
• Added internationalization support for multiple languages
• Fixed language selection to prevent reverting to English
• Removed language selection from user profile page
• Documented translation system architecture and key implementation rules
• Simplified language dropdown label from 'Translate summaries to' to 'Translate to'
• Updated custom voice count from 32 to 38 across all references
• Hid Eric, Lily, Laura, and George voices from profile voice selection
• Rearranged profile page layout and improved membership display
• Comprehensively fixed Profile page text readability in night-light theme
• Added text shadow effect to streak milestone text on Profile page
• Kept white text on Journey Beyond banner in night-light theme
• Made AnimatedStatCard component theme-aware for night-light theme
• Matched Library hover shadow to Community page style
• Fixed Library card rounded corners on hover
• Added and refined synthesis engine tooltip with triangulation terminology and instant show
• Enhanced admin chat functionality by adding secure authentication and access controls
• Updated dev-log subtitle to 'New features, improvements, and bug fixes'
• Added real-time support chat for CORE+ members
• Improved auto-scroll behavior in chat interfaces
• Added email rate limiting: only send notification on first unread message
• Updated support chat header and footer text
• Added support email mailto link to forgot-password page
• Updated support email text to 'with any questions'
• Updated maximum video length for premium tiers
• Added Dev Log page to show curated development timeline
• Updated Dev Log page to match Blog page's animated header style
• Adjusted page layout and text for optimal viewing on smaller screens
• Enhanced feedback widget with glass effect icon
• Show 'Manage CORE+' for premium members instead of hiding upgrade section
• Fixed CORE+ detection in Chrome extension
• Added secure Stripe payment information to CORE+ page
• Updated pricing tier description to reflect new benefits
• Updated wording on CORE+ page to emphasize support and access
• Updated username popup messages with clearer, low-stakes copy

2026-01-10

• Added advanced settings for synthesis engine selection (Claude Haiku 4.5 or Gemini Flash 3)
• Redesigned advanced settings UI to match new mockup
• Added retry logic to extraction API calls to prevent transient network failures
• Improved extraction engine descriptions and added tooltips
• Fixed error when changing language settings
• Updated styling to use theme-aware classes for better consistency
• Hid upgrade button for CORE+ members
• Updated Chrome extension manifest to match approved Chrome Web Store version
• Added CORE+ upgrade CTA for Deep Synthesis feature access
• Fixed Extract Wisdom button not appearing on YouTube videos
• Added smart token-based routing to prevent Claude rate limits
• Exempted CORE+ members from rate limiting
• Added welcome email system for new premium subscribers
• Added manual email sending capability for user communications
• Replaced Whisper/Supadata with Deepgram for all direct media file transcription
• Added CORE+ badge to user profiles for premium members
• Updated limit exceeded page to show CORE+ benefits
• Updated dashboard link to redirect to CORE+ management page
• Added language selector tooltip for guest users
• Doubled the size of the CoreWise logo on authentication page

2026-01-09

• Updated blog post content to reflect community interaction changes
• Updated blog post content for platform tips and tricks
• Added language selector to blog posts
• Fixed SVG icon display for blog posts
• Updated blog post display and enabled HTML rendering in markdown
• Added fallback images for blog posts
• Updated blog post thumbnails with new design and animations
• Showed video URLs and PDF source in AI errors tab
• Improved video synthesis process and error reporting
• Added content type to display uploaded PDF or video source
• Updated subtitle to accurately describe AI question-answering capabilities

2026-01-08

• Improved title extraction for TikTok and non-YouTube platforms
• Increased size of Mandala Fusion icon and its container
• Updated icons for deep synthesis feature to enhance visual appeal
• Added new icons for meta-analysis feature
• Updated capitalization of core features across the application
• Fixed missing videoUrl in apiUsage records
• Updated video length feature wording to be more descriptive
• Added extended video length extraction to premium features
• Updated pricing and feature descriptions on the CORE+ page
• Added error codes to extraction endpoints for better debugging

2026-01-07

• Added new blog post about information vs wisdom
• Improved how quotes are extracted for synthesis by adding new constraints
• Improved text cleaning by removing specific model tags and preserving formatting
• Improved AI synthesis prompt for handling nuanced insights
• Added aphoristic/distilled content handling to extraction prompts
• Updated extraction prompts with epistemic stance preservation
• Fixed Outside Perspective access control for CORE+ members
• Updated About page problem section copy
• Improved unique insights prompt clarity

2026-01-06

• Added infinite scroll to Library page
• Added Chrome extension link to navigation bar and updated button text
• Updated visual previews for extension popup and YouTube button
• Added interactive links to feature descriptions and titles
• Updated AI model name display
• Fixed infinite scroll with callback ref pattern
• Fixed community lazy loading
• Hid PDF upload toggle on mobile screens
• Reduced PDF upload size limit from 50MB to 25MB
• Added file signature validation to prevent non-PDF uploads

2026-01-05

• Added PDF upload and URL extraction with LlamaParse integration
• Added 500ms cooldown to footer hover particle effects
• Fixed mobile text overflow for rotating slogans
• Made device fingerprint optional for guest extractions
• Improved summary formatting by converting bullet points to headings
• Enabled blog post synchronization and markdown generation
• Updated AI model information in several blog posts
• Updated website's open graph image for better visual representation

2026-01-04

• Updated feature names and limits for premium tiers
• Added ability to download raw video and article transcripts
• Ensured unique model perspectives are accurate and tied to specific AI models
• Updated all user-facing references from GPT-4o-mini to GPT-5-mini
• Fixed GPT-5-mini temperature parameter issue
• Added OpenAI API parameters for speed and consistency optimization
• Improved loading indicators to show progress consistently
• Optimized GPT prompt for OpenAI automatic caching

2026-01-03

• Updated homepage rotating slogans
• Updated guided tour extraction step text
• Added magical hover effects with sparkling particles to navigation elements
• Added interactive particle text effects with accessibility improvements
• Increased upvote button clickable area on Community page
• Removed 'Synthesized Wisdom' label from extraction header
• Increased icon sizes for better visibility (delete, drag to collection)

2026-01-02

• Prevented users from submitting direct links to PDF files for extraction
• Fixed Notion export: properly extract text from quote objects
• Added canonical URLs to About, Blog, and individual blog posts
• Fixed SEO canonical URL issue - Community page now properly indexed
• Added font preload hints to reduce Inter font latency
• Fixed Chrome loading and thumbnail flickering with lazy loading and better fallbacks

2026-01-01

• Improved payment verification and checkout flow with test mode support
• Updated footer link to point to CORE+ instead of Community
• Ensured extraction progress display functions correctly
• Added export functionality to the prompt for new users
• Added interactive guided tour for new users with improved reliability

2025-12-31

• Created new baseline for database migrations and updated schema
• Updated extraction loading sequence visual checkmark timing
• Increased Deep Synthesis grid columns on larger screens
• Redesigned extraction previews to use header empty space efficiently
• Added extraction previews to completed meta-analyses
• Fixed Galaxy loading screen to only show for new meta-analysis creation
• Improved Deep Synthesis floating button UX and made helper text clickable
• Fixed tier recognition and enhanced Deep Synthesis Q&A context
• Added Q&A functionality and UX improvements to Deep Synthesis feature
• Fixed tooltip text visibility on collection buttons

2025-12-30

• Improved tier icons with tooltips, hover effects, and increased sizes
• Reduced WhimsyHeading font size to prevent letter cutoff on edges
• Added horizontal padding to Everything Unlocked heading
• Updated weekly streak description to be clearer about requirements
• Added synced pricing selectors at bottom of CorePlus page
• Applied Extract Wisdom styling to Everything Unlocked heading
• Fixed scroll glitchiness on CorePlus page with GPU acceleration

2025-12-29

• Enhanced profile page with larger icons, improved streak display, and cosmic portal
• Added 7-day milestone icon and removed streak flames
• Updated CORE+ upgrade messaging to highlight Audio/Podcast
• Updated Language Preferences description text
• Improved profile page layout with side-by-side sections and responsive fonts

2025-12-28

• Added showcase page for premium member custom icons
• Added animated title and highlight to the blog page
• Added podcast generation and playback with CORE+ tier gating
• Fixed Q&A scroll behavior to show start of AI responses
• Added correct thinking configuration for Gemini Flash 3

2025-12-27

• Improved voice filtering to show custom voices first and remove unwanted defaults
• Hid voice descriptions in profile settings, show names only
• Improved audio UX with in-browser playback and caching to save API credits
• Updated profile page to use new voice preference and preview icons
• Changed default voice to Daniel - Steady Broadcaster
• Added voice preview functionality to Profile settings

2025-12-26

• Improved extraction preview fallback with random preview hooks
• Updated AI summary storage to support raw text with backward compatibility
• Updated frontend to display raw markdown for source analysis
• Added extraction metrics table for enhanced logging and debugging
• Added support for H1 markdown headers in synthesis notes
• Added colorful gradient fallback for unavailable video thumbnails

2025-12-25

• Added bulk extraction feature with UI, API, and background processing
• Fixed Notion 2000 character limit by chunking long text
• Added Notion export troubleshooting guide

2025-12-24

• Updated homepage headings with larger special font and gold outline
• Enhanced homepage cards with layered shadows for better readability
• Added dedicated video thumbnail component for extraction pages
• Improved insight display by cleaning up extra brackets
• Added missing unique insights section to AI synthesis prompt
• Centered synthesis core with AI model loading animations

2025-12-23

• Added blog post: Claude Haiku vs Sonnet for transcript analysis

2025-12-22

• Implemented glassmorphism value prop blocks on homepage
• Added blog post: Claude vs GPT vs Gemini comparison
• Fixed video duration fetching and validation
• Added video length limit of 2.5 hours for all tiers

2025-12-21

• Disabled extract button when on CoreWise site
• Fixed Chrome extension authentication, popups, and loading issues
• Added Chrome extension for CoreWise
• Enhanced loading screen with convergence/divergence visualization
• Refactored EducationalOverlay to compact single-line badge

2025-12-20

• Updated Gemini model name display to Flash 3
• Ensured view counts on community and share pages show at least one
• Added text shadow to rotating homepage slogans
• Restored formatting and spacing for AI-generated text content
• Updated extraction page to display unique model perspectives with colored icons
• Improved extraction card previews with engaging hooks
• Improved how underlines are applied to headings for better readability
• Adjusted blog post background opacity for better glassmorphism effect

2025-12-19

• Updated homepage slogans to better reflect core value propositions
• Removed unused seeded persona system
• Updated text descriptions on the journey page for better clarity
• Ensured specific extraction labels are always bolded
• Added real-time search and filtering to community page with smooth animations
• Improved visibility of dropdown options when hovered over
• Added filtering and searching capabilities to the library page
• Improved application loading speed with code splitting and font optimization
• Improved language selector accessibility and simplified logic

2025-12-18

• Added ability to manage and disable publicly shared community content
• Fixed authentication card layout consistency
• Improved mobile layout to prevent horizontal scrolling
• Adjusted homepage heading font size for better readability on all screen sizes
• Updated welcome message and standardized authentication page width
• Increased AI model token limits and improved error handling
• Updated AI model prompts to focus on extracting wisdom
• Added streak protection to prevent streak resets with available protectors
• Fixed inability to click elements behind particle effect
• Fixed achievement and streak animations to prevent crashes
• Removed timed restriction on achievement celebrations

This dev log is updated regularly with user-facing changes. Committed to Improve.