Your Morning Brief Every Day

TL;DR: You can now schedule Alter actions and our new Morning Briefing action is a perfect fit. Now your start your day informed, prepared, and focused on what truly matters with summaries of your overnight emails and today’s schedule. And finally you can edit past messages in conversations, search in the action editor and enable smart pasting on background jobs.

New Features & Enhancements

• Morning Briefing: Added morning briefing action with scheduled hour support
• Dictation: Use realtime live transcription when available
• Background Jobs: Added system notification support and smart pasting capabilities
• Action Editor: Added search field and alphabetically sorted categories
• Conversation: Added copy, retry, and edit buttons for user messages
• Actions: Categorized default actions under Alter category
• Models: Gemini 3 Pro preview available for Pro & Lifetime subscribers

Bug Fixes & Stability

• Audio Stability: Fixed chipmunk voices, bird noises, choppiness, resolved crashes when switching processors (dictation > meeting > dictation), improved audio timing issues, and prevented crash when stopping mic+system
• Notch Input: Better support for Vietnamese input methods and fixed crash on Meeting Report source
• Chat & Conversation: Fixed crash when starting/closing a chat window, fixed escape key not stopping chat, fixed autoscroll when streaming starts, and fixed crash on saving message to conversation
• Action Editor: Fixed crash on parameter selection if index is out of bounds
• Workspace: Fixed search to be lowercase and remove punctuation, fixed renaming of workspaces, and fixed transcript context not being added to workspace
• Keyboard Shortcuts: Fixed crash in GlobalKeyboardShortcutListener
• Performance: Improved drop and context strip performance

Local+ Plans Fix

TL;DR: Fixed an issue where Local+ plan users were incorrectly treated as free users. All billing and plan recognition is now back to normal. Also includes image generation crash fix, improved speech-to-prompt handling.

New Features & Enhancements

• Speech-to-Prompt: Improved handling by avoiding cleanup for speech-to-prompt operations

Bug Fixes & Stability

• Critical Billing Fix: Resolved issue where Local+ plan users were incorrectly treated as free users - billing and plan recognition now working correctly
• Image Generation: Fixed crash when generating images
• Whisper: Fixed Whisper support for Perennial
• Tools: Fixed issue with listing quick actions
• Audio: Minor warning cleanup
• Typo: Fixed “allways” → “always” typo

Introducing Invisible Alter Actions

TL;DR: Background Jobs let you execute any Alter Action silently in the background without a conversation window. Perfect for grammar correction, customer support automation, and instant recommendations. Plus tons of fixes and new models including GPT-5.1, GLM 4.6, and Kimi K2 Thinking.

New Features & Enhancements

• Background Jobs: Enable background jobs for everyone to execute Alter Actions silently without a conversation window
• New Models: Added support for GPT-5.1, GLM 4.6, and Kimi K2 Thinking
• Editor: Can set temperature for actions
• Voice Triggers: Added flexible match at end for voice triggers
• Dictation & Speech: Disabled cleanup for speech-to-prompt, better cleanup prompt and a way to disable the cleanup completely
• Chat: No more auto resizing of the chat window and smoother auto scroll behavior when entering a new message
• Actions: Added “what to watch” action
• Push to Prompt: Display a notification when there is no text detected
• Billing: Can buy a plan from the website
• App Sense: Improved link serialization, and better regular window detection
• Tools: Improved Quick Actions for tool usage
• Settings: Separate blocks for local and local+ processors in the Voice tab
• Flow: Better tool lookup

Bug Fixes & Stability

• Crash Fixes: Resolved crashes in mic and system audio processor dealloc, mic indicator (NaN/infinity audio level values), model parameter view model deinit, tool gallery (empty path stack), application observer starting, action editor when changing actions, and during autoresizing of the chat panel
• Data Races: Fixed data races in installed apps, models, and active app name, pasteboard and audio.
• Notch: Fixed obnoxious notch input cursor position
• Threading: Enforced main thread while message streaming
• Library Updates: Updated tool client library to avoid crash on connection cancellation

Teach Alter Your Defaults

TL;DR: Control Alter settings directly from chat with the new Manage Settings tool. Update your profile, appearance, model parameters, and preferences without leaving your conversation. Plus Flow tool for multi-step tasks, improved onboarding, Apple login, and comprehensive stability fixes.

New Features & Enhancements

• Manage Settings Tool: Read and update Alter settings from any chat window, including profile, appearance, model parameters, notifications, and analytics preferences
• Scroll Tool: Added scroll tool with improved browser compatibility
• Get Active App Context Tool: New tool to retrieve context from the currently active application
• Wait Tool: Added wait tool for timing control in tool sequences
• Login With Apple: Added “login with Apple” support with improved webview experience
• Onboarding: Brand-new onboarding experience with tutorial and improved flow
• Action-Driven Dictation Cleanup: Dictation cleanup is now driven by an Alter Action for better customization
• Audio Ducking: More sensible audio ducking during dictation to improve quality
• Audio Processors: Use different audio processors for dictation and meetings

Bug Fixes & Stability

• Race Conditions: Fixed multiple race conditions
• Crash Fixes: Resolved crashes when restoring clipboard, capturing window images, listing windows, default audio processor, recording history, and analytics capture
• Notch: More reliable notch-closing behavior, fixed focus issues after selecting a model, improved hover delay, and better handling when clicking outside
• Voice: Fixed data race on keywords flag, improved transcript indexing, better error reporting for invalid audio files and removed hesitation parts of speech in English
• Audio Recording: Fixed echo in meetings, improved handling of low audio detection, and better error reporting for empty audio files
• Conversations: Fixed concurrency issues when renaming conversations and improved context handling
• IME: Fixed cursor position issues and input method suggestions
• API & Endpoints: Better error handling for API errors with body, fixed custom endpoints for Gemini, OpenAI, and OpenRouter
• Voice Triggers: Fixed automatic context addition issue
• JXA: Fixed bad access crashes

Introducing Live Notepad

TL;DR: Live Notepad turns passive live sessions into active intelligence. Real-time transcript + screenshots + questions + tool integrations = you play at a next level. Take notes, create tasks, build knowledge—all while staying present in the conversation.

New Features & Enhancements

• Live Notepad: Real-time AI assistant during meetings and videos with a toggle shortcut (default: cmd+shift+→)
• Screen Sharing Privacy: Option to hide Alter from screen sharing
• Screenshot in Conversations: Added screenshot capabilities directly within conversation windows

Bug Fixes & Stability

• Live Captions: Disabled for non-WhisperKit Pro users to prevent recording errors
• Conversation Renaming: Fixed issues when renaming conversations in the notch
• Conversation Window: Adding context now properly updates the context strip
• Transcript Actions: Fixed trash icon functionality in recent transcripts
• Audio Recording: Notification after 2 minutes of low audio detection
• Context Menu: Action line is now fully clickable
• Apple Calendar & Notes: Better descriptions for better accuracy at tool use

Performance Improvements

• Notch Opening: Minor performance improvements when opening the notch
• Recording Start: Reduced startup time when beginning a recording
• Main Thread: Avoided blocking main thread for more than 3 seconds when fetching keywords

Introducing Live Captions

TL;DR: You know those YouTube subtitles everyone loves? Now they work for every app on your Mac. Live Captions = real-time subtitles for meetings, podcasts, streams, everything. Auto-detects language, runs locally, fully customizable.

New Features & Enhancements

• Live Captions: Real-time transcription display during meetings with a toggle shortcut (default: cmd+shift+↓)
• Voice Recording: New shortcuts and menubar buttons for starting/stopping/cancelling recordings
• Voice Settings UI: Put shortcuts in their own window
• Models: Welcome to Claude Haiku 4.5

Bug Fixes & Stability

• Audio Processing: Fixed handling of large 2-channel 48000 kHz audio files
• Clipboard: Prevented interference with complex clipboard data
• Notch Behavior: Notch now closes when frontmost app changes away from Alter
• Tools: Connection to Google Apps was failing or timing out

3000 tokens/s, welcome Alter Fast

TL;DR: You can now use Alter Fast as a model router, a handpicked selection of the market’s fastest models. Be aware, speed is addictive

New Features & Enhancements

• Models: Introduced a fast Alter model pipeline with smart auto‑selection and 3000 tokens/second
• Notch Cursor: Added proper text field keyboard navigation in the notch
• Dictation: Now replacements respect word boundaries

Bug Fixes & Stability

• Notch & Windows: Kept Alter windows in front when closing the notch
• Infrastructure: Optimized load balancers configuration and grace periods during streaming on sigterm signals
• Conversations: Multiple crash fixes, fixed the invisible draggable area while minimized and better support of streaming finish reasons
• Models & Endpoints: Fixed custom endpoint where an empty URL produced an empty model list
• Audio & Recording: Mitigated a double‑free audio crash and the recording clock
• Core & Startup: Improved error reporting when Core Data fails and prevented crashes on MCP Client start errors

Action-Level Tool Customization

TL;DR: You can now choose exactly which tools each Alter Action can use, giving you surgical precision over your workflows. Define tools per action for faster execution, better accuracy, and context-aware automation. Plus dictation replacements.

New Features & Enhancements

• Browser Support: Added Wavebox to supported browsers
• Dictation Replacements: Apply custom replacements to dictation and speech to prompt, with new replacements section in the Voice settings
• Template System: Added {% userLanguage %} tag and {{ xxx | dominantLanguage }} filter for smarter language handling

Bug Fixes & Stability

• Conversations: Restored confirm deletion Enter key behavior
• Tool Management: Resolved crash with ghost tools, now processing tools in chunks of three to prevent freezing
• Editor: Prevented crash when selection ID is nil
• Diff View: Fixed magic diff toggle switch behavior
• Text Selection: Restored missing text selection in bubble view
• Speaker Identification: Improved speakerkit loading reliability
• API Layer: Isolated API on its own actor to resolve concurrency issues
• Core Data: Prevented indexing crash
• Audio Recording: Improved handling to avoid frame capacity crash when ending recordings

Alter + Pyannote 4

TL;DR: We just shipped Pyannote 4 / Community-1 — the latest open-source SOTA for speaker diarization, and we’re the first to bring it to Mac on day one. This means even better offline speaker identification for your local AI transcription.

New Features & Enhancements

• Speaker Identification: Ne state-of-the-art model pyannote4/Community-1
• Tool Management: Added ability to disable all tools for a given action

Bug Fixes & Stability

• Transcript Generation: Improved title generation for transcripts on Tahoe
• Audio Engine: Prevented crashes from “player did not see any IO” errors
• Clipboard Provider: Fixed crash when getting change count out of memory
• Conversation Window: Escape key no longer closes the conversation window, fixed crash on auto-resize
• Message Handling: Fixed retry functionality when last message is not an assistant message
• Focus Management: Possibly fixed focus issue after deleting last context in conversation
• Core Data: Resolved minor warning when loading conversations
• Thread Safety: Moved voice triggers setter to main thread, protected fetch keywords from re-entrancy

Voice Triggers are live

TL;DR: You can now trigger any Alter Action just by saying a magic word out loud. Say “Hey Charlie” to start a conversation with our Mental Model expert, or “Hey Google” to fire off a web search. Plus we added new Grok models, per-action tool selection, and squashed a ton of bugs!

New Features & Enhancements

• Voice Triggers for Alter Actions: Added voice triggers to quickly activate built-in and custom actions
• New Grok models: Added Grok-4-fast models
• Per-Action Tool Selection: Configure specific tools to use in actions for better accuracy and speed
• Title Generation: Utilize foundation models (Tahoe) for generating conversation titles
• Model Selection in Actions: Set “none” as default model to use the model selected in the notch
• Audio Processing: Audio files are now automatically treated as meeting content when dropped
• Action Editor: Enhanced picker behavior with consistent entry removal across interfaces

Bug Fixes & Stability

• Audio Crashes: Multiple critical fixes for audio engine crashes on device changes and buffer conversion
• Dictation: Fixed race conditions and double-free issues in dictation processing, restored 10-second minimum length for cleanup
• Conversation Management: Fixed message saving issues on load and auto-resize crashes
• UI Fixes: Resolved dark mode issues in tool selection, confirmation dialog destructive actions, and notch border appearance
• Markdown: Improved code block removal while preserving newlines and indentation, fixed diff view for grammar correction
• Memory: Removed NSImage from structs to prevent crashes and fixed constraint issues in floating windows
• Error Handling: Better error descriptions for WhisperKit failures and content extractor exceptions
• Dev Workflow: Updated workflow for Xcode 26 compatibility
• Performance: Debounced scroll content height monitoring for smoother conversation performance
• Sentry: Updated Sentry version and improved crash reporting initialization