Changelog

Restores universal media playback controls (play, pause, next, previous, seek) across all native macOS players and web media (Spotify, Apple Music, Netflix, YouTube, VLC, SoundCloud, etc.) via dynamic MediaRemote dispatch. Introduces automated high-resolution website favicon resolution and an impervious multi-tier artwork rendering engine that completely eliminates broken image placeholders from the Notch and SuperIsland.

Directly dispatches commands to the macOS MediaRemote subsystem (MRMediaRemoteSendCommand), enabling reliable play/pause toggling, track skipping, and seeking regardless of whether media is streaming in Chrome, Safari, Spotify, Apple Music, or VLC.

Introduces full autonomous multitasking for compound requests, allowing Motus to seamlessly execute multiple simultaneous and sequential tasks in a single instruction—such as cleaning the desktop, conducting deep web research, and saving synthesized briefs to Apple Notes—without dropped tasks or premature listening cutoffs.

Empowers Motus to handle compound multi-intent prompts across distinct system domains in a single turn, coordinating desktop file reorganization, autonomous web searches, and Apple Notes documentation seamlessly.

Delivers an instantaneous sub-second conversational intelligence engine with direct Tier-0 routing for everyday queries, automatic tool bypass for chit-chat dialogues, default-collapsed reasoning outlines for a clean chat feed, and the refreshed official Motus Pitch Deck V1 document integration.

Integrated ultra-fast instant matching for common greetings, status inquiries, system volume, media playback, and app controls, delivering natural conversational responses in under 2ms without LLM latency.

Major chat refinement introducing instantaneous execution with zero pre-flight AX tree bottlenecks, an on-brand spinning Motus logo during reasoning, inline task dropdowns with real website references, and hover copy/re-edit controls for user prompts.

Hovering over any sent message now displays one-click copy and instant pencil edit actions, allowing seamless prompt tweaks and fast re-submissions.

Critical stability release resolving chat prompt submission screen unmounting, calibrating inactive voice notch height to exact hardware dimensions across all Mac models, and integrating an application-wide ErrorBoundary for zero downtime.

Eliminated infinite re-render loops in markdown typewriter streaming and citation rendering, ensuring instantaneous response streaming with zero UI flickers or blank screen unmounts.

Delivers direct AWS S3 image and document storage with preview cards matching modern studio design, high-fidelity multimodal vision processing for AI coding from screenshots/mockups, auto-expanding scrollable chat textarea, and dark solid send buttons across all voice and input states.

Upload, paste from clipboard, or drag-and-drop images and documents with automatic upload to AWS S3, interactive preview cards with badges and removal buttons, and persistent rendering in conversation history.

Introduces an interactive model selector across both empty and active chats, high-precision native Whisper audio transcription with dedicated recording pill controls, automated split-screen live previews for generated web code & documents, and seamless instant upgrade verification.

Seamlessly choose from 5 AI intelligence tiers directly from the chat input and empty state: Serum 1.1 (Standard Fast), Serum 1.5 Pro (Balanced Reasoning), Aether 2.0 Ultra (Deep Coding & Logic), Omni 3.0 Vision (Multimodal & Screen Perception), and Titan 4.0 Max (Autonomous Intelligence).

Guarantees that all newly registered and pending verification users seamlessly progress through the interactive onboarding flow. Includes an elevated osascript fallback for seamless OTA auto-updates across root-owned installations.

Pending verification and un-onboarded accounts are now automatically routed to the step-by-step onboarding experience, preventing bypasses to the main chat interface until setup is finished.

Introduces a complete top-bar multi-model selector with credit-based tiering, inline thread renaming, continuous voice transcription accuracy fixes, and intelligent [Target Application] prompt routing. Extends the split-screen workspace to automatically preview all agent-generated HTML/CSS/SVG files instantly.

Select from 5 distinct intelligence tiers directly in the chat top bar: Serum 1.1 (Free), Serum 1.5 Pro, Aether 2.0 Ultra, Omni 3.0 Vision, and Titan 4.0 Max. Model choices instantly persist across sessions and route through the Motus credit gateway.

Completely hardens production network connectivity for all users and introduces a dedicated Direct S3 OTA background updater that bypasses macOS code signature download barriers. Delivers resilient API retries, timeout protections, and seamless background updates for all existing client installations.

Engineered a native Direct S3 updater daemon that polls manifest updates from AWS S3, streams delta binaries with background progress tracking, automatically cleans quarantine attributes (xattr -cr), and performs seamless 1-click in-place application restarts.

Transforms Motus into a complete multitasking operating environment. Introduces an integrated resizable split-screen workspace with a multi-tab in-app browser, instant live previews for generated websites, SVGs/banners, CSV data tables, LaTeX equations, and resumes, a bulletproof 4-tier filesystem deletion engine, permanent live API synchronization, and a native PKG installer with zero Gatekeeper friction.

Browse the web directly inside Motus without leaving your workspace. Features full multi-tab lifecycle management: open, switch, refresh, and close tabs alongside active AI agent threads with zero window clutter.

Replaces generic chat prompts with intelligent, semantic conversation titles generated by a dedicated AI title engine, enhances macOS accessibility tree traversal for deeply nested UI structures, and stabilizes mascot viewport zoom locking during rapid interactions.

Introduced a dedicated /ai/title endpoint that analyzes early conversation turns to generate concise, human-grade 2-4 word titles rather than echoing raw user prompts.

Major leap in autonomous screen interaction speed and accuracy: replaces slow cloud vision roundtrips with sub-25ms native Swift screen comprehension (Apple Vision OCR + AXUIElement accessibility tree), eliminates the legacy canvas third cursor, and delivers instant guaranteed click execution across macOS 14, 15, 26, 27 (Golden Gate), and 28.

Engineered a direct native bridge between the desktop client and MotusNative Swift daemon. Real-time Apple Vision OCR and AXUIElement tree traversal detect exact sub-pixel screen bounding boxes locally on Apple Silicon in under 25ms, bypassing multi-second cloud screenshot processing.

Introduced the precision secondary cursor companion—an autonomous floating vector pointer that glides alongside your workflow with zero container background. Features seamless system tray toggling, instant Clicky-style visual pointing, real-time mouse tracking, and expandable macOS Sequoia spatial chat.

Engineered a pure vector secondary cursor triangle with demure rim lighting and subtle drop shadow. Operates at the highest macOS window level with absolutely zero rectangular background container or white border artifacts.

Introduced the all-new liquid glassmorphism interface across both idle and expanded notch states with real-time backdrop blur and specular edge highlights. Rebuilt the voice recording capsule and quick chat to native Apple SF Pro typography and Control Center specifications, integrated fully interactive system settings with live visual feedback, and perfected continuous hands-free voice dialogue.

Engineered a native frosted glass rendering pipeline using backdrop-filter blur (48px), 240% saturation, and specular edge refraction. Glassmorphism now persists seamlessly into the idle/inactive notch wing geometry, blending effortlessly into any desktop wallpaper.

Released continuous voice conversation loops with seamless back-and-forth hands-free interaction, conversational voice cadence with relaxed silence detection, complete uncut statement delivery, a warm and natural human persona, real-time credit ledger spending integration, and automated OTA desktop updates.

Motus now remains in an active listening loop after completing responses or actions. You can hold a continuous back-and-forth dialogue without needing to manually click 'Talk Now' after every turn.

Introduced native Intel (x86_64) Mac support with Universal 2 binaries, live persistent media streaming with dual-state equalizers for Apple Music & Spotify, interactive spoken response scrolling with lyric-style highlighting, and high-speed AWS S3 distribution for direct downloads and silent OTA updates.

Motus now runs natively on both Intel (x86_64) and Apple Silicon (ARM64) Macs. All underlying Swift helpers and C++ addons have been re-engineered into Universal 2 binaries.

Completely redesigned the Voice Notch into the multi-tab Motus Island UI. Integrated real-time music rendering directly into the idle notch. Enhanced execution feedback with real-time progress indicators and human-like voice confirmations.

A complete redesign expanding the Voice Notch into a beautiful, multi-tab layout featuring Motus AI, a System Cockpit, an interactive File Shelf, Calendar, and Weather.

Reduced background memory footprint and optimized IPC message routing.

Optimized background memory usage for long-running sessions, reducing memory consumption by up to 35%.

Enhanced coordinate mapping to perfectly support multi-monitor desktop setups.

Enhanced screen coordinate mapping for multi-monitor setups, ensuring perfectly centered overlays.

Upgraded the internal Voice Activity Detection algorithm to dynamically calibrate out ambient background noise.

The microphone now dynamically calibrates to your environment's noise floor in the first 180ms.

Fine-tuned application drop shadows and glassmorphism elements to shine in macOS Dark Mode.

Improved vibrancy and glassmorphism contrasts specifically for dark mode environments to enhance text legibility.

Upgraded the underlying sorting algorithms for the Desktop File Organizer.

Upgraded the file sorting algorithm to categorize newly downloaded files intelligently by mapping extensions.

Dramatically improved the speed at which Motus becomes fully ready after launch.

Deferred loading for non-critical background modules to significantly speed up initial launch time.

Stabilized interactions with system-level applications like Apple Music.

Decreased latency when opening or focusing applications via voice commands.

Refined the background auto-update engine and networking resilience.

Refined the OTA update pipeline to ensure updates download cleanly in the background.

Shipped Desktop v0.2.0 release artifacts with silent background update checks, expanded multi-tier intelligence substrate routing (Luna, Terra, Sol), launched early access waitlist reservation flow, and updated transactional email templates.

Packaged Motus Desktop v0.2.0 DMG and ZIP distribution bundles with automated background update checks silent on startup.

Completed migration to official production infrastructure at heymotus.com, integrated direct Mac installer DMG downloads with Apple Silicon & Intel architecture selection, verified auto-updater release pipelines, and configured live Stripe billing fulfillment.

Connected landing page and navbar download actions directly to official GitHub releases with architecture detection for Apple Silicon (ARM64) and Intel (x64) Macs.

Architected unified Model Gateway with multi-tier routing (Luna, Terra, Sol), upgraded Voice Activity Detection for instant 1.2s transcription turnaround, added Whisper phonetic biasing, native Apple Reminders and Notes adapters, and dynamic multimodal vision decoupling.

Architected model-agnostic gateway separating logical tiers (Luna for conversational speed, Terra for balanced OS automation, Sol for frontier coding and reasoning) with prompt caching cost optimization and zero-latency failover.

Engineered native desktop file organization, added end-to-end messaging automation, enhanced voice speech pause tolerance to 2.2s, and upgraded background web research without popup windows.

Motus can now automatically scan all loose desktop items and sort them into categorized folders (Screenshots, Documents, Media & Videos, Installers & Archives, Code & Data) while keeping existing project directories untouched.

Shipped interactive product showcase experiences, multi-platform downloads with automatic architecture detection, and enhanced responsive navigation.

Experience live simulated multi-app workflows, screen vision understanding, and automated desktop tasks directly in the browser.

Integrated automatic background updates, fixed multi-virtual-desktop window visibility, and resolved authentication token loss during login handshakes.

Motus now checks for new releases automatically in the background and applies instant updates with a single click.

Implemented native application adapters, telemetry monitoring, and multi-tier execution recovery.

Built structured intent discovery for Slack, Spotify, Notes, Calendar, Finder, and Safari.

Introduced secondary Ghost Cursor for interactive UI element pointing, built extensive Settings panel, and unified product visual branding.

Created dedicated transparent overlay window with secondary ghost cursor that glides to exact coordinates to point out buttons, errors, or controls for the user.

Built native system binary for sub-100ms Spotlight file searches, added close app actions, and built administrative credit tooling.

Engineered standalone search binary leveraging Spotlight APIs to instantly search, reveal, and manage files from natural language.

Connected real-time screen awareness with vision routing, continuous multi-turn dialogue, and Stripe billing module with budget guardrails.

Added automatic desktop screen capture sent directly to vision models, allowing Motus to analyze code, UI errors, design tools, and windows.

Built full authentication infrastructure with Resend email OTP verification, session guardrails, and native Apple Calendar scheduling.

Engineered user registration, login, 6-digit OTP verification via Resend email service, and session persistence.

Major architectural leap to V2.0: Introduced physical MacBook hardware notch integration, native background daemons, global hotkeys, and fast-path system execution.

Created custom transparent window that fits exactly over the 14-inch/16-inch MacBook Pro hardware notch cutout with spring expansion animations, waveform visualizer, and menu bar pass-through.

Added explicit microphone access handling, updated sidebar theming, and refined chat attachment constraints.

Integrated macOS system media permission prompts before initializing audio recording contexts.

Added universal keyboard launcher accessible via Cmd+K to search conversations, trigger system actions, and switch views.

Built keyboard-driven Command Menu supporting fuzzy search over chats, settings, and direct OS action shortcuts.

Enhanced Chat component with smooth typewriter streaming animations, persistent thread renaming, and navigation badges.

Implemented character-by-character typewriter streaming for AI responses and automatic conversation title generation with persistent local storage.

Added Spotify and Apple Music playback automation and integrated full GitHub Flavored Markdown rendering with table support.

Built native playback bridge supporting search and immediate playback of tracks, albums, and playlists.

Generated high-resolution macOS application icons and updated visual branding across desktop views.

Built Apple iconset resolutions (16x16 up to 1024x1024@2x) and integrated icon assets into application bundle.

Initialized the Motus AI architecture: backend reasoning engine with IntentClassifier, speech-to-text processing, and desktop UI with action automation skeleton.

Scaffolded core reasoning architecture with model routing, audio buffer decoding, and task plan generation.