changelog what we shipped. no fluff, just features and fixes.
get notified when we ship
changelog delivered to your inbox. no spam, unsubscribe anytime.
v2.0.384 February 10, 2026 16 commits Speaker diarization improvements and critical CPU compatibility fixes
speaker diarization now uses advanced clustering with multiple strategies for better accuracy audio player now seeks directly to transcription timestamps for faster navigation windows installer visual design updated with new branding fixed crash on Lunar Lake CPUs by disabling AVX-512 in Windows builds fixed video frame seeking that caused audio/video misalignment fixed speaker identification in databases when using multiple devices full changelog on github
v2.0.383 February 10, 2026 12 commits Fixed video sync issues and improved timeline UI
timeline UI improvements for better navigation and usability video frame timing misalignment that caused audio/video desync during playback database speaker detection issues when using multiple devices full changelog on github
v2.0.343 February 6, 2026 10 commits Smarter monitor management and better transcription out of the box
Default transcription model upgraded to whisper-large-v3-turbo-quantized — significantly better accuracy with minimal speed trade-off Check for Updates option in the Apple menu for on-demand update checks Monitors that are plugged in automatically start recording, unplugged monitors stop cleanly — no restart needed Permission banner is now much more visible (solid red background) so you won't miss missing permissions Multi-monitor CPU usage optimized — lower overhead when recording multiple screens Fixed timeline getting stuck in an infinite loading loop when the server is unreachable Fixed chat streaming crash when canceling a response mid-stream Fixed noisy audio recovery warnings flooding error tracking (2,700+ false alerts eliminated) Added diagnostic breadcrumbs to tray icon operations for better crash reporting Fixed CLI builds failing on Linux and Windows full changelog on github
v2.0.312 February 5, 2026 30 commits Apple Intelligence integration and rock-solid macOS overlay
Apple Intelligence — on-device AI via Apple Foundation Models for local summarization and extraction new /v1/extract-todos endpoint with smart map-reduce chunking for large contexts OpenAI-compatible API server for Apple Foundation Models (use with any OpenAI client) app detection now uses the topmost visible window on your monitor instead of system-wide focus — no more wrong app labels when Finder briefly steals focus pinch-to-zoom works without clicking the timeline first search panel activates properly on macOS fullscreen Spaces eliminated overlay blinking during macOS Space transitions (7 iterations to get it right) fixed Tauri event listener race condition affecting 69 users (APP-2/5/9/W) localStorage access guarded for null WKWebView context (APP-4, 48 users) keyboard focus restored properly when showing/hiding overlay panels full changelog on github
v2.0.302 February 3, 2026 30 commits Arrow key navigation, search suggestions, and major reliability fixes
global arrow key navigation for timeline — press ← → to step frames, Alt+← → to jump between apps search modal now suggests interesting keywords from your recent activity pinch-to-zoom restored on timeline with zoom-level-scaled scroll speed URL detection precision improved from 84% to 95% using OCR garbage rejection heuristics app context popover now shows the full URL instead of just the hostname auto-advance past the status screen after 20 seconds unconditionally server now binds the HTTP port before reporting success — prevents silent "address in use" errors on restart recording restart no longer kills its own process when reclaiming the port timeline auto-refreshes stale data when window regains focus full changelog on github