Changelog
What we've shipped.
Recent improvements, fixes, and new features.
May 5, 2026
Document upload + topic picker
- newUpload PDF, DOCX, TXT, or Markdown files (up to 25 MB) — text is extracted automatically.
- newTopic picker: for content over ~500 words, pick which topics to keep before generating the script.
- newURL fetch now extracts article text upfront and surfaces topics for long pieces.
- improvedNew Episode dialog now asks which project to add the episode to, instead of guessing.
April 21, 2026
100 templates + studio previews
- newTemplate library expanded from 16 to 100 built-in templates across all major use cases.
- newEach template has a 30-second studio preview — actual voices and music, fully mixed.
- newMusic library expanded from 18 to 83 tracks (Incompetech CC-BY 4.0 + Pixabay).
- improvedVoice catalog upgraded to a faster TTS engine — better rate limits, snappier response.
April 14, 2026
Newsroom polish + RSS distribution
- newRSS feed generation per project — submit to Spotify, Apple Podcasts, Amazon Music.
- newPublic listening pages with embedded audio player and show notes.
- improvedNewsroom run detail page now shows full audit trail with retry button.
April 7, 2026
Production hardening
- improvedPoint-in-time recovery enabled across our entire database — 35-day rollback window.
- improvedAPI rate limiting (100 req/sec, 200 burst) and production alarms wired up.
- improvedFull observability for every AI call — script gen and agent mode traced for audit and debugging.
- fixedImage generation polling timeouts and 403 handling.
March 24, 2026
Agent Mode (premium)
- newChat-based assistant inside the workspace — ask in natural language, agent uses tools to get things done.
- newTool registry: list/get/create projects, episodes, personas, templates, plus search the catalogs.
- new30-day session retention with TTL.
March 21, 2026
Global CDN
- newAll media now served via time-limited signed URLs from a global CDN — faster everywhere, hotlink-protected.
- improvedOrigin-locked, signature-verified access for private media.
March 12, 2026
Async audio pipeline
- newParallel TTS pipeline — 30-segment episode generates in under 2 minutes.
- newRender engine with per-block mixing, background ducking, and transition stingers.
- newPolling UI shows segment-by-segment progress.