Changelog

A new overlay controls panel in the editor side panel lets you adjust fit mode, anchor position, crop, linking, and keyframes for any overlay you select. Linked overlays now also stay in sync automatically when you trim the parent clip, so you don’t have to readjust them by hand.

Agents

• Navigate between overlay keyframes with previous and next buttons in the new overlay controls panel.
• Annotations support time-range updates from the player position: start an annotation and it extends as the video plays.
• The crop modal supports an aspect mode toggle for free-form or locked aspect ratio cropping.
• Selecting words in the transcript shows the duration of the selection.
• Quality checks now warn when overlays drift out of sync with their clips.
• Mobile sharing of rendered videos now uses the native share menu.

App

• A new ”+ New Chat” button and a list of your starred and recent chats live in the side navigation, so you can jump into any chat without leaving the page.
• The search bar is properly visible against busy backgrounds and translucent when empty, with the blur effect now also working on Safari.

Project settings, feedback annotations, caption tools, and transcript editing have all moved into the editor’s side panel, so you can adjust your video without bouncing back to chat. The Podcasts beta also gains native sharing and preset caption styles for generated clips.

Agents

• Change your video’s aspect ratio from a new project settings panel.
• Add comments and annotations from a new feedback panel, with markers that persist across sessions.
• Browse and apply caption presets from a new caption tools panel.
• Transcript editing has moved into the editor side panel.
• Style presets now work for text overlays, not just captions, and you can describe how you’d like a preset customized before applying it.
• A “More Clips” button generates additional clips from the same podcast source without starting over.
• Generated podcast clips can be shared with a native share button, and you can pick from preset caption styles when creating them.
• Click any element in the viewport to select it directly, and press Escape to deselect.
• Smoother reframing transitions between clips when consecutive clips come from the same source.
• Quality checks now separate technical issues from creative suggestions, so it’s clearer what actually needs fixing.

App

• A new Podcast Clipper [BETA] entry in the side navigation opens your podcast projects directly.
• Workspace scrolling no longer freezes when the cursor sits over the Favorites row.

Turn any video or podcast episode into multiple ready-to-post short clips with one click. Upload a file, paste a URL, or pick from your library, and the AI finds the best moments and assembles the clips for you. The same week, the editor gains a transcript panel for editing the timeline directly through the words.

Agents

• Set up to 10 clips, with control over duration, aspect ratio, pacing, and reframing.
• Captions get custom styling, background music, and automatic transcript corrections.
• Track progress in real time as clips move through indexing, planning, and rendering.
• A new Podcasts button in the sidebar opens the feature.
• Transcript panel: drag to select words, edit or replace text, or delete timeline sections from the transcript.
• The current word highlights during playback, so you always know where you are.
• A quick-access transcript toggle sits in the editor toolbar.
• A “Proceed with Plan” button approves the assistant’s shared plan in one click.
• New model options: GPT-5.5, DeepSeek V4 Pro, DeepSeek V4 Flash, GPT-5.4 Mini, Kimi K2.6, and GLM 5.1.
• The assistant can set keyframe transition types when adding or updating keyframes on the timeline.

App

• A new Podcast Clipping section on the Workspace lets you start a podcast project, jump back into recent ones, and see them at a glance.
• A quick ”+” button on the Workspace’s Libraries section creates a new library in one click.
• Toasts now appear at the bottom of the screen.

Workflows

• Conditional task step: branches that run different tasks based on a variable’s value.
• Smoother step editor for long text fields, with a fullscreen mode for larger edits.

Pick a caption design from a live preview gallery before applying it. The reframe tool also gets vertical snapping, so the framing lines up without fiddling.

Agents

• Browse and preview caption styles, then apply the one you want.
• Reframe with vertical snapping to common positions.
• Project aspect ratio is now detected automatically from your clips.
• Undo and redo feel more responsive, and captions update after undo/redo.
• Cloned chats now appear in the chat list.

App

• New guided sign-up sets you up with a nickname and asks how you plan to use StarZero.
• Change your nickname anytime in Settings → Account.
• Visual and unified searches warn when your libraries don’t support Visual Intelligence, with a one-click switch to transcript search.
• Search results and library grids read top-to-bottom, left-to-right, so the most relevant items are upper-left.
• Hovering a library video no longer autoplays when “Autoplay videos” is off in Settings.
• The brief white flash when navigating is gone, and pages load faster.
• The mobile navbar and the desktop hover-mode search popover have a darker, more readable background.

Every render in your chat history is now a restore point. Click any past render to either start a new project from that moment or replace your current project’s state with it, timeline and all. A split render button also lets you export with or without a watermark in one click.

Agents

• Keyframe markers on the timeline now line up precisely with frame boundaries.
• Each project keeps its own undo/redo history when you swap between them, so switching back doesn’t lose your work.
• Session statistics show a credit breakdown by category with a visual chart.
• When reframing, the assistant can keep framing inside the source footage to avoid black bars.
• Better handling of clips with unusual frame rates.
• New chats no longer disappear if you briefly disconnect before sending.

App

• Top up credits or upgrade your plan directly from the upload screen, or from the error message when an upload fails for those reasons.
• Media imported from Frame.io shows the Frame.io icon, and can’t be added to the same library twice.

Workflows

• Instances no longer crash at startup when media is missing aspect ratio information.
• More reliable processing of queued instances.

Mobile

• Credit and storage information is visible when selecting files to upload.
• Buttons, menus, and drawers use a darker frosted-glass background for better readability.

Editing in the timeline now works without the mouse. Play and pause, step frame by frame, and jump between keyframes from the keyboard. Frame tick marks appear on the timeline when you zoom in far enough, and the playhead lands exactly on a frame when paused.

Agents

• A keyboard icon opens a reference of all shortcuts, grouped by category.
• Chat and editor panels are resizable; drag the divider, or collapse a panel entirely.
• Scrubbing is more precise, zooming no longer flickers, and the spot under your cursor stays put as you zoom.
• Rendered videos have a share button that copies a link to your clipboard.
• Playback speed control in the video player for rendered videos.
• The assistant re-reads the project timeline before making changes, so edits work from the latest state.
• Better assistant understanding of audio tracks, overlays, and how they relate to video clips.

App

• Frame.io links can be added via the upload modal.
• The library header shows upload progress with an estimated time remaining.
• Files upload 4 at a time instead of 2.
• Single-file uploads skip the selection screen and go straight to a summary with a preview.
• Login pages can be pre-filled with an email via a link, useful for redirects from verification emails.
• The verify-email page works without a session; you can resend the verification link from there.

Mobile

• Swipe navigation between chat and editor panels.
• Pinch-to-zoom on the timeline.
• Message input no longer auto-focuses, so the keyboard doesn’t pop up unexpectedly.

The app is now fully responsive. Browse libraries, search, upload, and manage your account from your phone.

Mobile

• Bottom navigation covering Workspace, Search, Libraries, Settings, and Agents.
• Vertical search scrolls like a feed and auto-plays the video closest to center; toggle with the lightning button.
• Video details open in a full-screen vertical layout with timeline, transcript, and download options.
• Full Account and Settings: profile, billing, API keys, password.
• Libraries get name filtering, sorting, grid and list views, and long-press to rename or delete.
• Uploads work end-to-end from your phone, with credit and storage warnings inline.
• Select and merge moments with a compact selection bar.
• Ask an agent directly from the library context menu.

Agents

• Copy and paste keyframes from one clip to another with Ctrl+C and Ctrl+V.
• Undo and redo in the editor with Ctrl+Z and Ctrl+Y (or Ctrl+Shift+Z).
• Timeline feedback: add comments and annotations directly on the timeline to mark specific moments.
• Playback speed control in the editor.
• Unlink audio and overlay clips from their parent video clip; clips stay linked through splits and merges.
• Pexels MCP: search for stock photos and videos to use in your projects.
• Better subject focus and narrative thread detection during automatic reframing.
• The assistant no longer asks for approval when working with content you’ve already selected.

Workflows

• Up to 20 clones per workflow, up from 10.

Recipes let you save step-by-step instructions once and reuse them across conversations, so the same multi-step task doesn’t need to be reassembled each time. They appear on the welcome screen for quick access.

Agents

• Crop overlays land directly in the editor with aspect ratio locking, for both video and image overlays.
• Split-screen tools redesigned with automatic subject detection for smarter layouts.
• Clone a chat in a new tab to branch a conversation into its own thread.
• Overlays can be linked to video clips, so they move and get removed together.
• Video segments preload when you open the editor, so playback is smoother once ready.
• Downloaded files keep their correct original filenames.
• More accurate cost estimates for image and video generation.
• Tool approval buttons are disabled when disconnected.

Ask Agents to generate images or short videos from a text description, optionally guided by reference images or frames from your video. Generated media comes with download, copy link, open in new tab, and click-to-view-full-size actions.

Agents

• Assistant shows estimated costs and asks for approval before generating.
• Auto-captions from text-to-speech are now timed word-by-word, so captions appear in sync with the spoken audio.
• Animated graphics can be placed directly on the timeline, moved, resized, and trimmed like video clips, and are included in rendered output and Premiere Pro exports.
• Add video and image clips from direct URLs to the timeline; trim, reposition, adjust audio, and animate them like library clips.
• Video feedback on rendered videos: draw, add comments at specific timestamps, and send feedback to the assistant.
• Split-screen rendering lets wide videos split top/bottom or left/right when rendering to a different aspect ratio.
• Editor undo/redo now supports up to 50 steps.
• Faster sidebar chat history search.
• Improved SVG graphics generation quality.

App

• Table view shows a YouTube ID column with filtering by full URL, ID, or partial match.
• Settings modal adapts better to smaller screens.

Workflows

• Shared templates can be shared without being listed publicly.
• Starter variables can be passed when starting a run.

Overlays are no longer a chat-only concept. You can now move, resize, and remove overlay clips directly on the timeline. Hold Shift while resizing to change speed instead of trimming, and overlay playback stays in sync with the main timeline during preview.

Agents

• Resize overlays in the video preview by dragging corner and edge handles.
• Drag overlays to reposition them in the preview; click to highlight for editing.
• Overlay keyframes: add, move, and delete position keyframes to animate overlay movement.
• Right-click overlay and audio clips in the timeline to quickly add keyframes.
• Overlay scale and fit options to choose how overlays fill their bounds.
• Overlays preload ahead of time for smoother playback transitions.
• Hovering the timeline scrub bar shows the current time position.
• Video streaming in the editor uses caching for faster loading when scrubbing.
• Custom MCP servers can return images, audio, and files; the assistant displays them in chat.
• Questions from the assistant support rich text formatting, with a Next button to skip between unanswered ones.
• Personalized greetings based on the time of day on the welcome screen.
• Saved recipes appear on the welcome screen for quick access.

App

• Click the logo to open the changelog.
• Libraries with a mix of YouTube imports and uploaded files show both Published and Added date columns.
• Grid cards use a uniform size based on the majority orientation in your library, with full-frame thumbnails.
• Search results no longer play audio unexpectedly; sound plays only on hover.

All renders now export at full resolution, a shift from the previous low-quality preview behavior. Free renders include a watermark; paid renders using credits remove it.

Agents

• Analysis results (charts, data, generated files) are saved permanently and can be shared via direct links or copied as artifact references for use in other chats.
• A new Relevant Artifacts section sits under analysis results with related data and files, plus download and copy buttons.
• The assistant can swap the audio source on an existing clip to a different library item or an external URL.
• Notifications close properly once all pending actions are approved.
• Premiere Pro exports include graphic overlays and external-URL audio clips.
• New voice browser for speech generation: listen and compare voices before picking one.
• Speech generation produces one sentence at a time for more natural delivery.
• Session statistics at the bottom of chat show token usage, credits spent, and tool counts.
• Audio clips support volume keyframes, visualized as a curve on the waveform and audible during playback.
• Render progress shows elapsed time, total duration, and estimated time remaining.
• Keyframes in the timeline show their type (ease, linear, hold) with distinct shapes; right-click to change.
• The assistant can ask clarifying questions before starting a task, with suggested options or free-text answers.
• OpenAI GPT-5.4 is selectable in Settings.

App

• New Workspace page collects your favorite libraries, recent libraries, and search in one place.
• Audio file uploads are now supported.
• Portrait (vertical) videos get their own column layout in the library grid.
• File metadata carries through from upload into your library.

The assistant can now layer picture-in-picture video, b-roll, image URLs, and generated graphics on top of your main video, with control over position, size, timing, and order. Overlays are managed through chat for now; editor support is on the way.

Agents

• The assistant can generate animated and static graphics (title cards, lower thirds, callouts) directly in chat, with interactive playback controls for animated ones.
• Generated graphics can use your video as a visual reference, producing titles and lower thirds that match the look of your content.
• Copy the URL of a generated graphic from chat.
• Image URLs (PNG, JPG, etc.) can be placed as overlays alongside video and graphics.
• More reliable playback of animated graphics in chat.

App

• API Keys: manage keys in Settings, with create, edit, rotate, and delete and scoped permissions.
• Pick an image file directly from the search bar.
• YouTube thumbnails and duration show immediately on cards while videos are still processing.
• Status labels are clearer: Queued means waiting to upload, Processing media means the server is working on it.
• Duplicate upload warnings shake to grab your attention if you try to add the same video twice.
• The upload modal no longer pre-selects all videos when adding URLs; pick which ones to import.

Ask the assistant for a soundtrack, intro jingle, or ambient bed and it will generate one. For more control, it can plan song structure, style, and lyrics before generating. Sound effects work the same way, with loopable output for ambient backgrounds.

Agents

• Separate vocals from background audio: the assistant can split a clip’s audio into voice and music tracks with independent volume control.
• Loop button on audio previews to repeat clips continuously in chat.
• /compact summarizes older messages in long conversations so the assistant keeps working without losing track of your goals, preferences, or decisions. Full chat history stays visible.
• /reload-settings applies your latest saved preferences to an existing chat without starting a new one.
• Exclude individual clips from captions when they already have burned-in subtitles.
• Browse and preview Epidemic Sound tracks directly in chat via MCP.
• Confirmation prompts before deleting messages or retrying from a previous point.
• See the inputs sent to external plugins via expandable details in tool results.
• Browser back and forward buttons now navigate between chats, and the tab title shows the current chat name.
• Claude Sonnet 4.6 and Gemini 3.1 Pro Preview are selectable in Settings.

App

• New Published date column in table view shows when YouTube videos were originally published.
• Redesigned upload modal with reorganized filters and sorting.
• Duplicate detection now works for file uploads, not just URLs.
• Open video details while a video is still processing; it updates live as the video becomes ready.
• Your preferred tab (visual or transcript) in video details is remembered across sessions.
• Clearing or retrying failed uploads is much faster.

Generated speech now supports emotion, reactions, pacing, and sound effects, for voiceovers that sound less flat.

Agents

• The assistant can set visual keyframes directly for more precise reframing control.
• More accurate video analysis with better understanding of timing and scene structure.
• Fixed reframing cases where the focus area could shift outside the video frame.
• Help documentation site at wiki.chat.starzero.ai with guides and tips for Agents.

App

• Facebook Ads URL support: paste a Facebook Ads Library link to import ads directly.

The assistant can now generate voiceovers from text using a wide selection of voices, with speed control from slowest to fastest. Audio can also be pulled from any URL (SoundCloud and similar) straight onto the timeline, without importing to your library first.

Agents

• Choose when the assistant reviews your projects: inline while editing, as a separate Task Agent step, or off entirely in Settings.
• Smarter automatic reframing can focus on specific subjects or objects when reframing to a different aspect ratio.
• Real-time render progress shows the export percentage as your video is processed.
• Project quality checks flag potential issues like flash frames, timing problems, or missing content.
• The assistant asks for confirmation before executing task lists.
• Save custom notes and metadata on library content for easier organization.
• More caption position options for higher or lower on-screen placement.
• Print analysis results directly from chat.
• Download interactive charts and widgets, or open them in a new tab for full-screen view.
• Scrub through video previews in chat by dragging the progress bar.
• Smoother timeline scrubbing and better performance when editing projects with many clips.
• Claude Opus 4.6 is selectable in Settings.

App

• URL unfurling: pasting a link detects and previews the video before you upload.
• Upload progress is smooth and continuous, with no more confusing “Retrieving video” messages.
• Bulk URL uploads are faster and show per-batch progress.
• Expired sessions redirect to login cleanly instead of showing confusing errors.
• Thumbnail previews work in Safari.

The app now warns you before uploading a video that is already in your library, so you do not burn credits and storage on content you already have.

App

• Failed uploads from one library no longer show up in another.
• Search keeps your library selection when navigating.
• Dragging files into the upload modal properly closes it after upload.
• Filters in table view work correctly with additive selections.
• Smoother transitions between the uploading and in-your-library states, with less flickering.

Agents

• More efficient analysis and processing of library content.
• Fixed a critical issue that could cause analysis tasks to produce incorrect results.

Failed uploads can now be retried directly from the card, without re-uploading. Library and media URLs are also human-readable now (e.g. /library/abc123-my-project), while old URLs still work.

Agents

• Library content shows published dates, which the assistant can use to filter and sort videos by when they were originally published.
• Generate captions from both video and audio clips, so on-screen text can pull from every audio source in your project.
• Print interactive charts and data visualizations directly from chat.
• Add, modify, and remove multiple custom text overlays at once.
• Clearer feedback when approving or rejecting assistant actions, with optional rejection reasons.
• More efficient library content analysis.

App

• MTS video files are accepted for upload.
• Watermark settings are remembered across sessions.
• Vertical (9:16) videos display correctly in the grid.
• Files and URLs are validated before uploading, catching issues earlier.

The assistant can now display interactive charts, data visualizations, and custom widgets inline in the conversation, so analysis results come to life as graphs rather than walls of text. To power this, it can also run code to analyze your data, perform calculations, and generate insights across your library.

Agents

• Fixed a rendering crash that could occur with certain keyframe configurations.

App

• The upload modal shows a summary of credit and storage costs before you confirm.
• Better progress indicators while videos are uploading and being indexed.
• URLs with special characters now work correctly.

Paste several links at once and import them in one go, instead of one at a time. The upload modal also gets quick filter chips for faster selection when there are many items to pick through.

App

• Faster and more reliable status updates while videos are processing.
• The app now only refreshes the videos you can see, making large libraries much faster.
• Credits are displayed in simpler numbers.
• Improved pre-upload status messages with more detail about what is being imported.
• Fixed purchasing credits not working.
• Fixed credits being displayed 1000x too large in the upload cost summary.
• Glossary editing in library settings (early preview).

You can now create, open, and switch between different video projects inside the same conversation, without leaving or cloning the chat. A project selector in the editor jumps between them directly.

Agents

• Quickly refine your project from the rendered video view with a new Refine in Editor button.
• More aspect ratio options when rendering.
• Improved web search results, with cleaner source citations and better-formatted answers.
• Thinking view auto-scrolls and auto-expands as new content appears.
• Enter inserts a new line in the message box. Use Cmd/Ctrl+Enter to send.
• Notifications when the assistant needs approval for tool calls.
• New caption font sizes (small, normal, large) and a background-highlight caption style that adds a subtle backdrop behind each word.
• Gemini 3.0 Flash Preview is selectable in Settings.

App

• Refreshed library page design with consistent card styles and correct file size display across views.
• Rename, edit, and delete a library directly from its page.

You will now get a browser notification when the assistant finishes a long-running task, so you do not have to keep the tab in focus to know when your render or delegation is done. A spinning favicon also shows activity at a glance while the assistant is working.

Agents

• Edit chat titles directly from the chat header by clicking the title.
• Progress indicators and a details button show what the assistant is doing during delegated tasks.
• Delegation tasks show more detailed stats and clearer status information.
• Smoother playback controls and more responsive keyframe editing.
• Render is now called Export in the editor.
• Settings has a save button with clear saved-state feedback.
• MCP Servers is now called External Plugins.
• Thumbnails pause generating during video playback for better performance on slower devices.

App

• YouTube Shorts are supported: import Shorts and filter them in your library.
• Sorting and filtering stay in sync when switching between grid and table views.

The editor timeline now has dedicated undo and redo buttons so you can step backward or forward through your changes. Add and remove keyframes at any point in your clips, and hold Ctrl (or Cmd) while resizing to snap the zoom level to common aspect ratios for precise framing.

Agents

• Timeline scrubbing works on touch devices, with playback stability improvements on phones and tablets.
• Video mute preference persists across sessions.
• First and last keyframes are more distinguishable on the timeline.
• Fixed fullscreen mode on mobile so controls appear properly.

Workflows

• Bulk instance creation shows a clearer preview table with instance numbers, media counts, and filenames.
• Video statistics appear alongside your media, showing feedback metrics and instance counts.
• Filter library and video statistics by workflow template.
• Approval rate metric shows how many videos received positive feedback at a glance.
• Steps editor uses collapsible sections, with short hints showing step details at a glance.
• Save button renamed to Launch with a rocket icon.

Define your own step-by-step workflows with descriptions and save them alongside the assistant’s built-in ones, so it can follow a process tailored to how you work. Settings is redesigned with organized tabs for agent preferences, external tools, and custom recipes.

Agents

• Visual keyframe editing on the timeline: markers show where reframing changes happen, navigate with skip buttons, drag to reposition the focal point, and resize the zoom by dragging corner handles. Touch support included.
• Find exact moments in your videos by searching for specific actions or events within a time range (e.g. “when the handshake happens”).
• Detailed descriptions of any section of your videos up to 60 seconds long, for better context.
• Control whether reframing can zoom out to show more of the scene or keep the original framing size.
• The assistant automatically merges consecutive clips that are adjacent in the source video, keeping your timeline cleaner.
• Smarter project planning using library statistics.
• Output videos now use a blurred background instead of solid black.
• 2:3 and 3:2 aspect ratios are available when rendering.
• Editor appears in the sidebar for easier side-by-side viewing with the conversation.
• More reliable video thumbnails and cleaner tables in chat.
• Fixed a crash when exporting project files for brand-new projects.

Workflows

• Better handling of task interruptions when stopping tasks.
• Fixed completed tasks being incorrectly cancelled by budget checks.

Reframe your videos to any aspect ratio (landscape to portrait, square, custom) while keeping the important parts in view. The assistant automatically adjusts the visible area throughout your video, following the action and keeping subjects centered, with a tracking dot that shows where the camera focus is moving in real time.

Agents

• Viewport outline can be toggled for better visibility when reframing, so you see the exact boundaries of your frame.
• Zooming out while reframing now snaps to common aspect ratios.
• Split clips at any point on the timeline, merge consecutive clips, and remove a section of time with everything in that range automatically cleaned up.
• Task lists support adding new tasks and updating existing ones for better project management.
• A new recipe creates standalone social media cutdowns from long-form source videos like YouTube uploads.
• Content summaries give quick overviews of your video clips.
• Map-reduce operations process large libraries in batches and combine results.
• Version notifications when a new version of the app is available.
• Clearer approval messages when the assistant requests powerful tools.
• Gemini 3.0 Pro Preview is selectable in Settings.

Workflows

• Gemini 3.0 Pro Preview is available in Settings.
• Better task handling during system maintenance.

App

• The upload modal warns you if adding videos would exceed your storage limit.

Track progress on multi-step work with a checklist view showing what’s done and what’s next. The assistant can also follow recipes, pre-defined workflows that guide it through complex tasks step-by-step.

Agents

• Type @ in the message box to reference video content naturally, or # to search your library for clips.
• Unified search experience in /search.
• Audio clips now export to Premiere Pro with full support in project files.
• Link and unlink audio clips from video clips for more flexible editing.
• Videos render at maximum resolution.
• Clear the entire timeline at once with a single command.
• Enhanced message input with smarter suggestions and a better layout.
• Welcome screen adapts to mobile and smaller screens.
• Structured fields in task planning for clearer instructions.
• Clearer error messages, with a dedicated last-error button in logs.
• Tables in chat get sticky headers and better row visibility.
• Content terms like Source, Extract, and Cut-down are clearly defined for easier understanding.
• Video rendering is more reliable, with automatic retries when issues occur.

App

• Drag-and-drop file upload works on grid view.
• Fixed a crash when deleting libraries.

Audio is now a first-class citizen in the timeline. Add audio clips separately from video, layer multiple audio sources, adjust each clip’s volume and speed independently, and see waveforms inline so you can edit by ear and eye at once.

Agents

• Add floating audio clips that can overlap in the timeline, giving more creative control over audio layering.
• Detach audio from a video clip to edit it separately and mute the original video.
• Ctrl+Wheel zooms in and out on the timeline for precise editing.
• Shift+Drag audio clip edges to adjust their speed.
• Volume warnings appear when audio clips are set above 100%.
• Drag and drop audio clips between tracks with visual feedback.
• Smarter transcript corrections with a dedicated assistant that fixes mistakes more precisely.

App

• Open in Chat button on video and moment cards jumps straight into a conversation about any video.
• New file upload flow with a preview modal and upload summary before confirming.
• Custom library thumbnails: set any image as your library’s cover.
• Speaker labels now use initials and distinct colors, making it easier to tell speakers apart in transcripts.
• Updated library card design with gradient and improved hover effects.

Workflows

• Rate workflow video outputs with thumbs up/down feedback directly from the video list.
• Dedicated feedback pages let you navigate between videos and provide feedback on each.
• Filter video outputs by workflow template.
• Pagination for large lists of outputs.

Search your conversations from the sidebar, with ranked results that are easy to scan. Render videos directly from the editor with a dedicated button, and download finished renders with their correct filenames.

Agents

• Right-to-left language support in chat and Markdown.
• Predictions and chat titles match your conversation language.
• Clickable clip and content labels in chat, with previews on hover.
• Captions and clip timing are validated to prevent impossible timestamps.
• Task Agent handles critique messages better, resulting in fewer failed tasks.
• Smarter retries when something goes wrong.

Workflows

• Template AI assistant now uses GPT-5 and shows its reasoning.
• Clearer guidance in the Task step on how critiques and scores are applied.
• Clone step count is validated with clearer limits and messages.
• Credit usage displays are more accurate, including cached token savings.
• Reliable video downloads with correct filenames on instance pages.

App

• Cancel button in table view selections.
• Empty libraries and filtered-out results show helpful overlay messages instead of a blank page.

View your plan, manage your subscription, and purchase credits directly from new billing and pricing pages, so you do not need to leave the app to handle account changes.

Agents

• On-screen text styles and fonts look better, with more consistent sizing.
• Thumbnails load more reliably during editing.
• Captions render at a better vertical position.

App

• Drag-and-drop image search: drop an image onto the search bar to find visually similar moments.
• Bulk delete: select multiple videos and remove them at once.
• Opening a video from table view updates the URL for easy sharing.
• Middle-click a library card to open it in a new tab.
• Watermark toggle lets you choose whether to include a watermark when rendering.
• Upload status shows accurate progress instead of getting stuck at 50%.
• Add all videos now asks for confirmation before uploading.

Workflows

• Sign in with your identity provider (OIDC).
• Clone templates and run them directly from the template pages.
• Undo/redo keyboard shortcuts work consistently on Mac and Windows/Linux.

Switch your library between grid and table layouts, with sortable columns for the larger libraries where thumbnails alone are not enough. Bulk instance creation in Workflows also lands this week, letting you spin up many instances at once.

Agents

• Plan tasks in a dedicated modal with clearer guidance.
• Clone chats from the sidebar.
• Smarter clip boundary handling with clearer warnings; buffers only apply when clips align to word boundaries.
• More accurate flash-frame detection when crossing shots.
• More reliable stopping and fewer stuck sessions.

App

• Chat button is disabled while a video is still processing, with a tooltip explaining why.
• Cancelling an upload properly removes the media entry instead of leaving a ghost card.
• Moment thumbnails load via a faster preview API.
• Fixed table view requiring a double-click to open a video.

Workflows

• Bulk instance creation: create many instances at once, choosing how to distribute media.
• Library management with size and aspect insights, including a width-height heatmap.
• Clone step supports a dynamic count based on a variable or expression.
• Cancel running workflow instances, with clearer status updates.
• Budget management and credits tracking for steps and tasks.
• Task step editor supports structured and freeform task inputs.

Ask Agents to fix mistakes in the transcript and in on-screen captions, so misheard names and wrong words do not end up baked into your output.

Agents

• Task agent supports optional rules to guide its behavior, with a clearer settings UI.
• Gemini Flash and Gemini Flash Lite are selectable in Settings.
• Inline videos in chat look cleaner.
• Fixed a rare crash when opening a new chat.
• Clearer errors and more resilient sessions.

Workflows

• Callback URLs are supported and notifications include more details.
• Run results include video URLs when available.
• Task rules can be added in the steps editor to guide the assistant.
• Session IDs are included in variables and results for easier tracking.

App

• Mute and autoplay preferences for moments are remembered between sessions.

A major editor upgrade lands with a new timeline, drag-and-drop clips, a preview player with preloading, and an on-demand thumbnail generator. Click the timeline to seek and adjust volume directly.

Agents

• Smarter zoom controls and cleaner typography.
• Faster video loads on slow connections and smarter thumbnail generation.
• More stable playback and synchronization, with more reliable Safari HLS playback.
• Rendering is more reliable for long jobs.
• YOLO mode in Settings bypasses approval prompts for MCP servers.
• Clearer confirmations before the assistant takes actions, with pending actions shown more plainly.

App

• Autoplay toggle for moments: when enabled, visible moments play silently and unmute on hover.
• Video details modal shows a loading skeleton while content loads.

Browse your library from inside a conversation, with previews and quick filters so you can find the clip you want without leaving the chat.

Agents

• Untitled chats are hidden from the list for a cleaner view.
• Simpler, clearer settings for how the assistant reviews content and uses context.
• Server logs in chat are ordered by time and easier to read.
• Better video rendering options and more consistent video height.
• Clearer guidance during video critique.
• More reliable tools with clearer error messages.
• Smoother video downloads and more reliable video handling.

App

• Opening a video updates the URL, so you can share direct links to a video’s visual or transcript tab.
• Improved keyboard navigation with visible focus outlines and auto-focused input fields.

Workflows

• Choose the AI model directly when setting up assistant settings.
• Smoother login experience.
• Clearer guidance when configuring assistant settings in the steps editor.
• More accurate cost tracking.
• More reliable instance pages after sign-in.

The upload modal now handles more than 100 videos in a single batch, so large YouTube playlists and channel imports no longer need to be split up. The modal also flags which videos are already in your library before you confirm.

Agents

• More stable conversations with smoother restarts and clearer errors.
• Credit usage is tracked more accurately.
• Content selection and syncing in chat is more reliable.
• Cleaner, easier-to-read text in the sidebar and chat.
• Video messages look better in chat.

Workflows

• Switch templates mid-workflow.
• Start a workflow with a set of starting choices.
• Move steps up and down in the editor for easier reordering.
• Simpler setup of assistant settings using templates.
• Smarter template checks and clearer guidance while editing.
• Numbered lists and text formatting look more natural in the editor.
• More reliable updates between the editor and running chats.
• Easier editing of settings in popups.

Workflows is a new surface for running and managing automated video pipelines at scale. Sign in, edit templates, and manage running instances from a dedicated app, with a Markdown editor that supports variable previews and clearer formatting.

Agents

• Control which messages are visible in chats.
• Clearer download filenames for videos and project files.
• Settings button moved in the sidebar for easier access.
• Voice recording stops cleanly when disconnecting.
• Credit balance displays correctly in edge cases.
• The assistant is more consistent at following earlier instructions.
• Refreshed visuals and branding.

App

• Live chat support via Intercom, reachable directly from the app.
• Upload cards show upload speed and estimated time remaining.
• Deselect all button replaces the close button when selecting videos or moments.
• Fixed the default library loading forever on new accounts.

Workflows

• Faster, more reliable workflow runs and chat session handling.
• Smoother template editing, including archiving.
• Track expected results and credit usage for runs.
• Cleaner instance pages.

Narrow a moment search to a single video instead of the whole library, so you can find specific shots inside long uploads. Transcripts can also now be downloaded as SRT subtitle files for use outside StarZero.

Agents

• Simpler app layout that feels faster and cleaner, with an ALPHA label in the sidebar and easier-to-find settings.
• Chats feel more responsive during longer thinking moments.

App

• Playlist and channel uploads show a confirmation modal with a summary before importing.
• Fixed vertical videos not displaying correctly in the player.
• Fixed shot descriptions and transcripts not loading in the merge modal.

Search across multiple libraries at once, instead of one at a time. The upload modal also gets a playlist picker, so you can choose which videos from a YouTube playlist or channel to import rather than adding them all.

Other changes

• Credit balance is visible in the sidebar.
• Change your password from Settings.
• Redesigned upload modal with section headers and faster scrolling for large lists.
• Clear all filters button in the upload modal.
• New-user welcome dialog.
• Fixed downloaded files having incorrect filenames.

The app detects and skips files you have already uploaded, so repeat imports do not double up on storage.

Other changes

• Open original button on video cards jumps back to the YouTube, TikTok, or other source URL.
• Blurred preview thumbnails appear while the full thumbnail is loading.

A new search experience brings everything together under one surface, replacing the scattered search paths from earlier versions.

Other changes

• Improved library switcher with search, better handling of long names, and fixed gradients.
• Large file downloads no longer freeze the browser.
• File uploads no longer freeze the interface when uploading many files at once.
• Highlighted transcript phrases show gradients for approximate timestamps.
• Horizontal sections scroll with the mouse wheel by default.
• Fixed library sorting not working correctly.

Videos in search results now prioritize what you can actually see, so visible results load first and scroll-heavy sessions feel more responsive.

Other changes

• Gradient overlays appear when scrolling through moments and library content.

Preview your merged moments with full playback controls directly inside the merge view, so you can check a cut before committing to it.

Other changes

• Search URLs remember whether you searched by visual or transcript.

Select multiple moments, drag to reorder, and merge them together into a single clip. No more copying timecodes between tools to stitch search results.

Other changes

• Play videos directly on the library page via a modal.

Trying to register with an email that already has an account now redirects you to the login screen with the email pre-filled, instead of dead-ending with a generic error.

Filter your libraries by name, and search media by title within a library, so you can find what you need without scrolling through a long grid.

A new timeline lets you navigate through moments visually, so finding the right shot is no longer a scroll-and-hunt job. Drag and drop files anywhere on the page to upload (not just in the upload box), and retry failed media directly from its card.

Other changes

• New video player controls: seek to a timecode, scrub with arrow keys, play and pause with space.
• Getting started guide for new users to help with first-time setup.
• Delete failed media directly from the status overlay on the card.

Click any moment to find visually similar ones across your library, so you can gather matching shots without having to describe them in words.

Other changes

• Pasting a YouTube playlist or channel link now asks what you want to do instead of guessing.
• Real-time status indicators when entering URLs: spinners while checking, green checkmarks for valid URLs, and clear error messages.

See every video currently being processed and any that have failed in one place, with options to retry. No more guessing which file got stuck where.

Other changes

• Password strength criteria is shown during registration.
• Demo library gets a badge and a custom thumbnail.
• Favicon adapts to dark and light mode.

Paste a YouTube channel URL to import videos in bulk, not just single videos or playlists. New users can also sign up through a dedicated registration form.

Other changes

• Download moments directly from the video details modal.
• Library page shows loading skeletons while stats are loading.

Find visually similar moments across your library, so you can discover related shots without needing to describe them. A new video details modal also lands, giving you visual and transcript views for any video in your library.

Other changes

• Privacy Policy and Terms and Conditions pages.

You can now delete libraries you own (except the demo and default libraries), so unused workspaces do not have to stick around forever.

Right-click any video to jump straight into a conversation about it, so starting an Agents session from the library takes one click instead of two surfaces.

Other changes

• Updated sidebar design with smoother open and close transitions.
• Videos on the library page wait for scrolling to stop before loading, reducing lag.

A new library content header lets you sort your media by the columns that matter, instead of taking the default order as-is.