An AI video studio for course and knowledge creators
Turn one lecture
into 30 days of short videos
Send us a course recording, a talk or a livestream replay. Within 72 hours you get ready-to-post clips with covers and copy for YouTube (long-form and Shorts), TikTok, Instagram, X, Facebook, LinkedIn, Reddit, Xiaohongshu, Douyin, WeChat Channels, Bilibili and more. AI does the editing, automatic checks catch problems, and a person reviews anything they flag.
- You only review the clips our checks flag
- Every clip comes with a cover, title, caption and tags
- We never log in to your accounts; you press publish
How it works
One lecture in, a whole batch out
-
Upload
Send one course recording, talk or livestream replay (ideally 90 minutes or less) and tell us which platforms and style you want.
-
Batch edit
AI splits the recording by topic and removes pauses, filler words and repeats. Then it adds captions, title bands, code zooms and notes panels across the whole batch.
-
Automatic QC
Every file is checked for loudness, audio/video sync, missing caption words, text hidden under app buttons, clip length and title length. Anything that fails is marked red.
-
Review exceptions
A person watches, fixes and re-renders every red clip, plus a random sample of green ones. You don’t have to watch them all.
-
Publish packages
You get one folder per platform with videos, covers, titles, captions, tags and a reminder to add the AI-content label. You do the posting.
- YouTube 16:9 + Shorts 9:16
- TikTok 9:16
- Instagram Reels 9:16 + feed 4:5
- X 16:9 / 1:1 / 9:16
- Facebook Reels 9:16 + feed 4:5
- LinkedIn 16:9 / 1:1
- Threads 9:16
- Reddit 16:9
- Pinterest 9:16
- Snapchat Spotlight 9:16
- Xiaohongshu 3:4
- Douyin 9:16
- WeChat Channels 9:16
- Bilibili 16:9
- Kuaishou 9:16
- Weibo 16:9
- Zhihu 16:9
- Dailymotion 16:9
- Kwai 9:16
Real numbers
One real batch: what worked and what didn’t
These numbers are from an internal test batch on one of our own 72-minute lectures, not a customer project.
- finished files
- 96
- 24 clips × 4 platforms, each with a cover and post copy
- AI API cost for the batch
- $0.73
- About $0.03 per clip; transcription ran locally
- clips passed automatic QC
- 21 / 24
- All 3 red clips were at edit seams and went to a person
- first-run machine time
- 3–4 h
- After a fix, re-running just the 13 affected clips took 18 minutes
What isn’t good enough yet
- Chinese captions still get some words wrong. A glossary fixes some of them and a person catches the rest.
- Filler-word cuts still need too many human yes/no calls (about 15 per clip in this batch). We’re working on letting the low-risk ones through automatically.
- When on-screen text in the source is small, it’s hard to read after a vertical crop.
What you get
What you receive
-
Per-platform videos
YouTube long-form 16:9 and Shorts 9:16; TikTok, Instagram and Facebook Reels, Snapchat and Pinterest 9:16; X, LinkedIn and Reddit in the shape closest to your video; Xiaohongshu 3:4; Douyin, WeChat Channels and Kuaishou 9:16; Bilibili, Weibo, Zhihu and Dailymotion 16:9. Each one is exported separately at that platform’s loudness level, with copy that follows its rules (title, length, hashtags, links), in English when your video is in English.
-
Captions
Burned-in captions that stay clear of each app’s buttons and title area. SRT files on request.
-
Covers
One per clip, sized and cropped for each platform’s feed.
-
Post copy
A title, caption and tags for every clip, within each platform’s limits and ready for you to edit.
-
QC report
Results for every file: what passed, why anything was marked red, and what the reviewer did about it.
-
Publishing checklist
A suggested posting schedule and a reminder to tick the AI-content label on each platform.
Design-partner program
Free for the first 30
We’re just getting started and need honest feedback and real data. The first 30 knowledge creators get one batch free. In return, you help us find out whether this is worth paying for.
You get
- 1 recording → 10 short clips
- 1–2 platforms
- Cover and post copy for each
- Delivery in 72 hours
In return, you agree in writing to
- Let us show the batch as a case study (it can be anonymous).
- Share views and saves 7 days after posting (screenshots are fine).
- Do one 15-minute feedback call.
- Agree up front on a price for a second batch if you’re happy, for example ¥499 for 20 clips. You can still say no.
Course and knowledge creators with recordings or livestream replays come first, then podcasters and bilingual creators. Spots are given by fit and order of application, and we reply to everyone.
Apply as a design partner
The online form is not open yet
For now, please apply by email with the details below. We reply to everyone.
Apply by emailhello@example.com
Pricing
Early pricing Early pricing
We’re still testing these prices with pilot customers, so they will change. Design partners pay whatever second-batch price they agreed to in writing.
-
Pilot
¥499 / batch
- 1 recording (up to 90 min)
- 20 clips × 2 platforms
- Covers and post copy
- Delivery in 72 hours
-
Monthly
¥2,999–4,999 / month
- 4 recordings a month
- 80–100 clips
- 2–3 platforms
- We keep your style and glossary
-
Per clip
¥25–40 / clip
- Extra clips beyond the monthly plan, or one-off jobs
- Same automatic QC and human review
Prices are in CNY. Outside China? Email us for a quote.
Desktop app (preview)
Prefer to do it yourself?
The same engine also runs as a desktop app on your own computer. Create batches, track them on a board, review clips in a grid and cut by editing the transcript. When it’s time to post, it fills in each platform’s upload page in a built-in browser, and you click publish yourself.
This is a preview build, so features and installation will change. If the download page has no files yet, the preview isn’t open; apply as a design partner instead. You’ll need your own AI API keys.
Open source
The editing engine is open source
Under the hood is video-studio, an MIT-licensed Claude Code skill. Describe the edit in English or Chinese. It covers 12 workflows, including talking-head shorts, lecture slices, podcast clips with face masking, explainers and vlogs, plus covers, captions, loudness and multi-platform export.
Each workflow was tested on real footage. Known issues are listed in VALIDATION.md in the repo.
View on GitHub · github.com/zyziyun/daycut# Claude Code skill · MIT
git clone https://github.com/zyziyun/daycut \
~/.claude/skills/video-studio
~/.claude/skills/video-studio/install.shFAQ
Privacy, rights and platform rules
If I use the studio service, where does my footage go?
You send it to us the way we agree on (for example a cloud-drive link) and we process it on our own computers, not on a cloud server. The video and audio files are not sent to AI services. To pick segments, write titles and proofread captions, the transcript text, a few keyframe screenshots and titles are sent to AI providers’ APIs (currently Anthropic and OpenAI). Transcription runs locally by default; audio goes to OpenAI for transcription only if local transcription fails and you agree.
What about the desktop app or the open-source skill?
Your footage stays on your own computer by default and never reaches us. As above, transcripts, keyframes and titles go to the AI services you configure, using your own API keys; AI voiceover or AI video generation sends text and reference images to that vendor. The MediaPipe component used by the engine may send anonymous usage statistics to Google; we will explain this, and how to turn it off, inside the app.
What do the AI providers do with that data?
It is handled under each provider’s API data policy, which decides retention and whether it is used for training. The privacy policy links to those policies so you can check them yourself.
Do I need to label AI content when posting?
YouTube, TikTok, Instagram, Facebook, Xiaohongshu, Douyin, WeChat Channels, Bilibili and others have disclosure or label rules for AI-generated or AI-edited content; follow each platform’s current rules. Our publishing checklist reminds you to tick the AI-content label by default. We do not log in to your accounts or post for you.
What about other people in my videos (guests, students)?
You confirm you have their consent to use their face and voice. Where you do not, we can cover faces with stickers, blur name labels, change students’ voices, or leave that part out.
When is my data deleted?
By default, your source footage and intermediate files are deleted within 30 days of delivery. Email us any time to delete earlier; we act within 7 days and confirm by reply. Data already sent to AI providers follows their retention policies, and we cannot delete it on their behalf.
When do the 72 hours start?
Once all the footage has arrived and the brief is confirmed. If we are going to be late, we tell you in advance.
Who owns the videos?
You own your footage and the finished clips. Case studies are shown only as far as you agreed in writing, and can be anonymous.
Contact
Questions? Send us an email
Partnerships, questions and data-deletion requests all go to this address. For the open-source skill, you can also open a GitHub issue.