I had Claude make the same video twice, 5 months apart. The difference is staggering.
Back in May, I had Claude Code make a demo video for Videowright, my open source tool for making videos in Claude Code. Last week I had it make the same video again: same script, same brief, but Opus 5.5 instead of Opus 4.7. The attached video compares them.
The video gen tool: scosman/videowright
What changed in 5 months
- 🎨 Design: May was basic text on a grid. Now it's rich animation and vector graphics, with a logo that draws itself.
- 🌍 3D: May was a spinning polygon. Now it renders real-time scenes on the GPU (the planet's coastlines, clouds and atmosphere are computed per pixel).
- 🖱️ Product UI: May was a dashboard that played itself. Now someone's actually using it: a cursor, a panning camera, a real route change.
- 🗣️ Narration sync: May synced by line. Now it's timed to every word.
- 🎙️ Voice: May was a stock ElevenLabs v2 voice. Now it's my own cloned voice on ElevenLabs v4, which can take emotion cues.
It's a fair comparison
Same script, same brief. Claude built the new one from scratch, and never opened the old video's code or renders. The style brief was literally "rebuild the style with the new model making the decisions".
It wasn't a one-shot. My notes were mostly taste: "too much text on screen", "more breathing room here", "this title is too busy". Claude did the rest: code, motion, audio sync, mixing.
The two full videos
Old video (May): on YouTube
New video (Nov): on YouTube
This video was made the same way
The comparison video above was also made in Claude Code with Videowright. It plays both real renders inside the motion graphics, syncs every cut to my voice, and mixes in music and sound effects.
What Videowright adds
While you can just prompt Claude Code to make a video, Videowright adds a bunch of helpful features:
- Live preview: a dev server that hot-reloads as you build. Say "make that motion a bit slower" and see it immediately.
- Styles: built-in templates, or create your own to reuse across videos
- AI voiceovers (ElevenLabs, Gemini, others), or record your own: it gets word timings and syncs the video to your pace
- Sound effects and music: sources and mixes many audio sources. Gen AI and open-source options.
- Mix in real video: drop in MP4s and it works them into the motion design
- Export to MP4: deterministic frame-by-frame rendering, no dropped frames
It's free and open source: scosman/videowright
What should I compare next? Hopefully folks find it helpful!
2 comments
Yes! I got annoyed that I had to manually close old finder/preview windows, so I OSSd an app that closes any stale window. There have been commercial apps in this space a decade, but none worked how I wanted (per window not per app, config each app separately). So I just made it with Claude using some spare tokens at the end of the week. You can customize anything to be the way you like it now.
scosman/Squeegee
G: Meet Biscotti: a free, private meeting recorder for macOS. It creates speaker-labeled transcripts, summaries, and action items—all processed locally on your Mac. No cloud, meeting bots, accounts, or subscriptions.
scosman/Biscotti
3 views
1: Biscotti is a free, open-source macOS meeting recorder that creates speaker-labeled transcripts, summaries, and action items—all processed locally. It works across meeting apps without bots, subscriptions, or accounts, keeping your audio and notes pri… scosman/Biscotti
5 views
2: Biscotti is a free, native macOS meeting recorder that keeps everything local. It captures audio from any app without joining calls, then creates speaker-labeled transcripts, summaries, and action items—no account or subscription required. scosman/Biscotti
4 views
3: Meet Biscotti: a free, private Mac meeting recorder that transcribes and summarizes locally—no cloud, bots, accounts, or subscription. Works across meeting apps and even in person. scosman/Biscotti scosman/Biscotti
2 views
I’m vibecoding an excel in rust just to see if it can be done and what are the limits.
scosman/freecell
So far it’s GPU accelerated (GPUI), cross platform, supports 95% of excel functions, supports charts, is about 40x faster than Numbers, and can render 240fps.
I thought I’d leave this process with a new appreciation for manual dev/details. I haven’t found that yet. I’m getting more into architecture details, writing more benchmarks, testing more. I know there are limits, but I’m still trying to find them.
It would be absolute garbage if I wasn’t already an engineer. It probably is absolute garbage in ways I don’t know because I haven’t read the code. But the app is… quite good. If someone invested into something like this full-time for 6 months it's hard to imagine what you'd need excel for.
There is only one truly AI native browser: scosman/cursed_browser
Biscotti - 私密会议录音原生 macOS 工具
可生成带说话人标注的转写与摘要
👉 scosman/Biscotti t.co
2 likes · 1 repost · 319 views
Biscotti v1.0: a free meeting recorder for Mac that never uploads your audio. Now recognizes voices and lets you chat with your meetings (MCP)
Problem
I'm tired of meeting transcription apps that send your private audio to their servers, aren't Mac native, and require an account/subscription.
I'm an ex-Apple engineer who built Biscotti: a private and free meeting recorder for macOS. Recording, transcription, and AI all run on your Mac, so nothing is ever uploaded. Free, no account, no subscription, no upsell, no data collection, native SwiftUI.
v1.0 is out! Since my first post here it's grown from 45 to 330 GitHub stars.
Download: scosman/Biscotti
New since the last post
MCP, language detection, and import all came from requests in the last threads on reddit, and five of you sent PRs. Thank you!
- 🗣️ Recognizes voices - learns your voice and the people you meet with regularly, and tags them automatically. It matches voiceprints from your past meetings, plus clues like "Hi, I'm Steve", and your calendar. All local.
- 💬 Chat with your meetings (MCP) - ask Claude or other MCP apps things like "What did Sarah and I agree to last time?"
- 🌍 Detects language - detects the spoken language instead of assuming English.
- 🔍 Instant search - full-text search across 5,000 meetings in under 1ms.
- 📥 Import / export - bring your notes over from CSV, including Granola exports.
- 📖 Custom vocabulary - teach it names and jargon, or pull them from your calendar automatically.
- Plus: more reliable when switching mics/speakers, better Teams detection, and lots of fixes.
Comparison
Lots of apps in this space. But Biscotti is the only one that's free (no upsell), Local AI, with Mac native UI. Others are:
- Cloud AI (your audio or transcript leaves your Mac), paid tiers: Granola, Otter, Fireflies, Fathom, tl;dv, Jamie, Fellow, Read AI, Notion AI Meeting Notes
- Local AI, closed source, paid tiers: Talat, TranscribeX, Alter
- Local AI, source available, not Mac native (web app wrappers), telemetry or a cloud upsell: Anarlog, Meetily
- Free, Mac-native UI, local AI, zero data collection, source available: Only Biscotti (that I know of)
Switching from Granola? v1.0 can import your existing meetings.
Pricing
100% free, no upsells, no paid features, source available on GitHub.
New to Biscotti? What Biscotti does
- 🔒 Private by design - your audio, transcripts, and notes never leave your Mac.
- 🧠 AI summaries - automatic summaries, action items, and meeting titles.
- 🤖 No bots, any app - Zoom, Teams, Meet, FaceTime, Slack huddles, or in person. Nothing joins your call.
- 📅 Calendar-aware - offers to record upcoming meetings, and stops automatically when the call ends.
- 🎤 Voice isolation - records your mic and everyone else as separate, clean channels. No echo.
Private AI == Local AI
Every model runs on your Mac: transcription, speaker ID, and summaries. It works fully offline once the models are downloaded. Your meetings live on your machine, and you own and control them.
I spent years at Apple building Photos, Memories, the lock screen, and more, and I wanted this to feel like it belongs on the Mac: native Swift/SwiftUI, launches instantly, small on disk. Not an Electron app or a port.
Requirements: Apple Silicon (M1+), macOS 15+, 16GB RAM recommended.
App details: Notarized, Free (no upsell), Private (zero data collection, all AI is local), Source Available on GitHub (PolyForm Perimeter License), 250 GitHub stars, AI used in development (but not vibe-slop)
About me: the subreddit rules ask for my LinkedIn: linkedin.com . Biscotti webpage: biscottiapp.com
What should I build next? And if you still use a cloud note-taker, what's keeping you there?
23 points · 29 comments
Mentions
I had Claude make the same video twice, 5 months apart. The difference is staggering. Back in May, I had Claude Code make a demo video for Videowright, my open source tool for making videos in Claude Code. Last week I had it make the same video again: same script, same brief, but Opus 5.5 instead of Opus 4.7. The attached video compares them. The video gen tool: scosman/videowright What changed in 5 months - 🎨 Design: May was basic text on a grid. Now it's rich animation and vector graphics, with a logo that draws itself. - 🌍 3D: May was a spinning polygon. Now it renders real-time scenes on the GPU (the planet's coastlines, clouds and atmosphere are computed per pixel). - 🖱️ Product UI: May was a dashboard that played itself. Now someone's actually using it: a cursor, a panning camera, a real route change. - 🗣️ Narration sync: May synced by line. Now it's timed to every word. - 🎙️ Voice: May was a stock ElevenLabs v2 voice. Now it's my own cloned voice on ElevenLabs v4, which can take emotion cues. It's a fair comparison Same script, same brief. Claude built the new one from scratch, and never opened the old video's code or renders. The style brief was literally "rebuild the style with the new model making the decisions". It wasn't a one-shot. My notes were mostly taste: "too much text on screen", "more breathing room here", "this title is too busy". Claude did the rest: code, motion, audio sync, mixing. The two full videos Old video (May): on YouTube New video (Nov): on YouTube This video was made the same way The comparison video above was also made in Claude Code with Videowright. It plays both real renders inside the motion graphics, syncs every cut to my voice, and mixes in music and sound effects. What Videowright adds While you can just prompt Claude Code to make a video, Videowright adds a bunch of helpful features: - Live preview: a dev server that hot-reloads as you build. Say "make that motion a bit slower" and see it immediately. - Styles: built-in templates, or create your own to reuse across videos - AI voiceovers (ElevenLabs, Gemini, others), or record your own: it gets word timings and syncs the video to your pace - Sound effects and music: sources and mixes many audio sources. Gen AI and open-source options. - Mix in real video: drop in MP4s and it works them into the motion design - Export to MP4: deterministic frame-by-frame rendering, no dropped frames It's free and open source: scosman/videowright What should I compare next? Hopefully folks find it helpful!
Yes! I got annoyed that I had to manually close old finder/preview windows, so I OSSd an app that closes any stale window. There have been commercial apps in this space a decade, but none worked how I wanted (per window not per app, config each app separately). So I just made it with Claude using some spare tokens at the end of the week. You can customize anything to be the way you like it now. scosman/Squeegee
G: Meet Biscotti: a free, private meeting recorder for macOS. It creates speaker-labeled transcripts, summaries, and action items—all processed locally on your Mac. No cloud, meeting bots, accounts, or subscriptions. scosman/Biscotti
1: Biscotti is a free, open-source macOS meeting recorder that creates speaker-labeled transcripts, summaries, and action items—all processed locally. It works across meeting apps without bots, subscriptions, or accounts, keeping your audio and notes pri… scosman/Biscotti
2: Biscotti is a free, native macOS meeting recorder that keeps everything local. It captures audio from any app without joining calls, then creates speaker-labeled transcripts, summaries, and action items—no account or subscription required. scosman/Biscotti
3: Meet Biscotti: a free, private Mac meeting recorder that transcribes and summarizes locally—no cloud, bots, accounts, or subscription. Works across meeting apps and even in person. scosman/Biscotti scosman/Biscotti
I’m vibecoding an excel in rust just to see if it can be done and what are the limits. scosman/freecell So far it’s GPU accelerated (GPUI), cross platform, supports 95% of excel functions, supports charts, is about 40x faster than Numbers, and can render 240fps. I thought I’d leave this process with a new appreciation for manual dev/details. I haven’t found that yet. I’m getting more into architecture details, writing more benchmarks, testing more. I know there are limits, but I’m still trying to find them. It would be absolute garbage if I wasn’t already an engineer. It probably is absolute garbage in ways I don’t know because I haven’t read the code. But the app is… quite good. If someone invested into something like this full-time for 6 months it's hard to imagine what you'd need excel for.
There is only one truly AI native browser: scosman/cursed_browser
Biscotti - 私密会议录音原生 macOS 工具 可生成带说话人标注的转写与摘要 👉 scosman/Biscotti t.co
Biscotti v1.0: a free meeting recorder for Mac that never uploads your audio. Now recognizes voices and lets you chat with your meetings (MCP) Problem I'm tired of meeting transcription apps that send your private audio to their servers, aren't Mac native, and require an account/subscription. I'm an ex-Apple engineer who built Biscotti: a private and free meeting recorder for macOS. Recording, transcription, and AI all run on your Mac, so nothing is ever uploaded. Free, no account, no subscription, no upsell, no data collection, native SwiftUI. v1.0 is out! Since my first post here it's grown from 45 to 330 GitHub stars. Download: scosman/Biscotti New since the last post MCP, language detection, and import all came from requests in the last threads on reddit, and five of you sent PRs. Thank you! - 🗣️ Recognizes voices - learns your voice and the people you meet with regularly, and tags them automatically. It matches voiceprints from your past meetings, plus clues like "Hi, I'm Steve", and your calendar. All local. - 💬 Chat with your meetings (MCP) - ask Claude or other MCP apps things like "What did Sarah and I agree to last time?" - 🌍 Detects language - detects the spoken language instead of assuming English. - 🔍 Instant search - full-text search across 5,000 meetings in under 1ms. - 📥 Import / export - bring your notes over from CSV, including Granola exports. - 📖 Custom vocabulary - teach it names and jargon, or pull them from your calendar automatically. - Plus: more reliable when switching mics/speakers, better Teams detection, and lots of fixes. Comparison Lots of apps in this space. But Biscotti is the only one that's free (no upsell), Local AI, with Mac native UI. Others are: - Cloud AI (your audio or transcript leaves your Mac), paid tiers: Granola, Otter, Fireflies, Fathom, tl;dv, Jamie, Fellow, Read AI, Notion AI Meeting Notes - Local AI, closed source, paid tiers: Talat, TranscribeX, Alter - Local AI, source available, not Mac native (web app wrappers), telemetry or a cloud upsell: Anarlog, Meetily - Free, Mac-native UI, local AI, zero data collection, source available: Only Biscotti (that I know of) Switching from Granola? v1.0 can import your existing meetings. Pricing 100% free, no upsells, no paid features, source available on GitHub. New to Biscotti? What Biscotti does - 🔒 Private by design - your audio, transcripts, and notes never leave your Mac. - 🧠 AI summaries - automatic summaries, action items, and meeting titles. - 🤖 No bots, any app - Zoom, Teams, Meet, FaceTime, Slack huddles, or in person. Nothing joins your call. - 📅 Calendar-aware - offers to record upcoming meetings, and stops automatically when the call ends. - 🎤 Voice isolation - records your mic and everyone else as separate, clean channels. No echo. Private AI == Local AI Every model runs on your Mac: transcription, speaker ID, and summaries. It works fully offline once the models are downloaded. Your meetings live on your machine, and you own and control them. I spent years at Apple building Photos, Memories, the lock screen, and more, and I wanted this to feel like it belongs on the Mac: native Swift/SwiftUI, launches instantly, small on disk. Not an Electron app or a port. Requirements: Apple Silicon (M1+), macOS 15+, 16GB RAM recommended. App details: Notarized, Free (no upsell), Private (zero data collection, all AI is local), Source Available on GitHub (PolyForm Perimeter License), 250 GitHub stars, AI used in development (but not vibe-slop) About me: the subreddit rules ask for my LinkedIn: linkedin.com . Biscotti webpage: biscottiapp.com What should I build next? And if you still use a cloud note-taker, what's keeping you there?