🏢 Hangzhou Gaishi Technology Co., Ltd. · National High-Tech Enterprise

Say anything.
Get it in one click.

An all-in-one AI platform for talking-head video — write, shoot, perform, edit. From a single idea to a video you can publish: script, teleprompter, on-camera delivery and editing, all inside one app.

Visit the site See the core capabilities
Shuode AI talking-head creation
5M+
active users
a category leader in China
4.9 / 5
across app stores
30,000+ ratings
27 languages
multilingual AI avatar
world's first
14 tools
AI features covering
write · shoot · perform · edit
POSITIONING

One app for the whole pipeline of a talking-head video

Most tools solve one link in the chain: the scriptwriter ignores shooting, the camera app ignores editing, the avatar tool ignores the teleprompter. Shuode AI puts writing, shooting, performing and editing into a single workflow — no exporting and re-importing between tools. The gain comes from the pipeline itself.

✍️

Write · AI script

Give it a topic and a tone and get a full script; rewrite drafts, break down viral scripts, and pull the text out of any video, audio file, image or document.

🎬

Shoot · AI teleprompter

A prompter that follows your speaking pace instead of scrolling at a fixed speed, plus virtual backgrounds, AR actors, beauty presets and sentence-level retakes.

🧑‍💼

Perform · AI avatar

Don't want to be on camera? Use a 1:1 clone of yourself, a ready-made model from the public library, or a still photo brought to life.

✂️

Edit · smart cut

Filler words, pauses and stumbles are detected and removed automatically; subtitles, denoising and watermark removal come built in.

THE STRONGEST PIECE

The AI teleprompter: it follows your pace, not a fixed speed

This is the capability Shuode AI is best known for, and the reason millions of creators keep it on their phone. A traditional prompter scrolls at a constant speed, so you end up serving the machine. Ours listens and moves with you.

A traditional, constant-speed prompter

  • The text scrolls at a fixed rate — you have to match the machine
  • Pause, ad-lib or slow down and the script runs away from you
  • Delivery sounds like reading, because your attention is on keeping up
  • One stumble means re-shooting the whole take

Shuode AI: speed-following

  • Speech recognition tracks where you are and scrolls to match your pace
  • Pause or improvise freely — the script waits for you
  • Delivery stays natural, because you are talking rather than chasing text
  • Sentence-level retake: redo one line instead of the whole video

Everything around the shoot, in the same app

  • Floating prompter: sits on top of any other app — live streaming, video calls, online interviews — and you can switch scripts mid-stream.
  • Pro hardware support: DSLRs, Bluetooth remotes, phone remotes, selfie sticks and gimbals.
  • Sentence-level retake: step back line by line; pause anywhere to change person, outfit or background and continue — the segments join seamlessly.
  • Virtual background: real-time matting without a green screen, or professional keying with one; 1,000+ backgrounds; works with the major streaming platforms.
Shooting with the AI teleprompter

Real footage

Twenty teleprompters compared
Playable on the hosted version
20 teleprompters, compared
An independent creator's verdict after testing more than twenty phone teleprompters.
Prompter and smart cut in practice
Playable on the hosted version
Prompter + smart cut
A real creator taking one talking-head video from script to finished cut.
AI avatar output
Playable on the hosted version
AI avatar output
What the avatar actually produces when you never step in front of a camera.
WORLD FIRST

Multilingual avatar: one script, 27 languages

On 25 September 2025, Shuode AI V6.0 shipped the world's first multilingual AI avatar — video generation and video translation fused into one step, turning a single-language script into avatar-delivered videos in Chinese, English, French, German, Japanese and 22 more.

The usual approach: multilingual dubbing

  • The footage is generated first, then another language is laid over it
  • Languages differ in pace and sentence length, so audio and picture drift apart
  • Lip shape, expression and gestures are still locked to the original language's rhythm
  • Every language needs manual clean-up, so volume breaks the process

Shuode AI: multilingual native generation

  • Translate the script first, then generate the picture — the order is reversed, and that is the point
  • Lip shape, expression and body language are generated per language, to its own pace and sentence length
  • Every version is a native one: audio and picture in sync, delivery natural
  • Post-production review costs collapse, so going multilingual actually scales
STEP 1

Pick an avatar

Your own custom clone, or a ready-made character from the public library.

STEP 2

Enter one script

Write it once, in one language. No need to prepare 27 translations.

STEP 3

Tick the languages

Chinese, English, French, German, Japanese… 27 in total, multi-select.

STEP 4

Generate

Finished multilingual videos come out directly — generation and translation in one pass.

Zhejiang Radio report: a Yiwu merchant selling pearls in 27 languages
REAL CASE

A Yiwu merchant selling pearls in 27 languages

"Based in Yiwu, selling pearls worldwide." A merchant in Yiwu — China's largest small-commodity hub — used the multilingual avatar to turn one piece of sales copy into 27 language versions and reach buyers across the globe. The story was covered by Zhejiang Radio, and the opening of the Yiwu Global Digital Trade Center made CCTV's national evening news.

  • One script covers 27 languages — marketing reaches overseas markets directly
  • No hunting for a presenter, voice actor or reshoot per language — export content finally scales
  • The same capability serves cross-border e-commerce, foreign-trade sites and overseas social media
FASTEST-GROWING USE CASE

Commerce avatar: photograph the product — the model, the set, the story and the selling points are filled in for you

All you do is photograph the goods. Who presents them, where, how, and which selling points to hit — the AI fills all of that in and returns a finished selling video with the product in hand. This is not 24/7 unattended live-streaming; it compresses "cast a model → build a set → write a script → shoot → edit" into a single upload.

🧑‍🎤

The presenter

Hundreds of ready-made models covering Chinese and international looks, genders, ages and styles; or define your own brand spokesperson from a prompt or a reference photo, or use your own cloned avatar.

🏙️

The setting

The AI matches scene and camera to the product's style. The product is genuinely integrated — held, worn or put on the body — rather than pasted on top of the frame.

✍️

The story

No idea how to write selling copy? The AI writes it. You can also supply your own by text or audio. Hundreds of voices across ages, genders, styles and dialects.

🎯

The selling points

Expression and gesture follow the copy, which is why it reads as real rather than as a lip-synced avatar; multi-angle cuts highlight the details that convert.

Shooting a selling video the old way

  • Find a model who both looks right and can present — the hardest step
  • Then schedule, build the set, do hair and make-up, light it and shoot it
  • Nobody on the team knows how to write copy that converts
  • No editor in-house, so it goes to an agency; the smaller your volume, the worse the unit cost

With the Shuode AI commerce avatar

  • Upload the product photo, pick a model or let the AI generate one
  • The AI casts, writes and edits — no live shoot required
  • Style is matched automatically; held, worn or dressed all integrate
  • Half an hour covers what used to take half a month, with no dependence on anyone's calendar
Commerce avatar in use
LATEST RELEASE

Commerce avatar 2.0: presentation plus multi-angle AI editing

No more single fixed camera. On top of the avatar's delivery, 2.0 layers multi-angle, multi-camera AI editing that shows the product's detail and how it looks in use — visibly more information per second, and more persuasion. On your side, it is still one product photo.

  • Held products: electronics, food and drink, toys and collectibles, baby and pet, auto parts…
  • Worn accessories: hats, glasses, bags, scarves, jewellery, watches…
  • Apparel: clothing and footwear, swapped directly onto the model
  • Commerce videos output in six languages (Chinese, English, Spanish (MX), Japanese, Korean, Indonesian); for 27-language export volume, pair it with the multilingual avatar above

Real cases: one product photo → one finished video

Each of these was given nothing but a single product photograph. The model, the setting, the gestures and the spoken copy are all generated.

Insulated tumbler — generated selling video
Playable on the hosted version
Tumbler product photothe uploaded photo →
the AI generated the video
Insulated tumbler
A still product shot becomes a presenter holding it and talking about it.
Lipstick — generated selling video
Playable on the hosted version
Lipstick product photothe uploaded photo →
the AI generated the video
Lipstick
A close-up shot becomes a swatch demo, with gestures following the copy.
Hairbrush — generated selling video
Playable on the hosted version
Hairbrush product photothe uploaded photo →
the AI generated the video
Hairbrush
A plain white-background photo turns into a scene with an actual explanation.
Beef crisps — generated selling video
Playable on the hosted version
Beef crisps product photothe uploaded photo →
the AI generated the video
Beef crisps
Flat packaging art becomes a tasting segment at a table, product and person integrated.

More finished output

Commerce avatar output
Playable on the hosted version
Commerce avatar output
What comes back after a product photo goes in: model, setting and copy all generated.
Commerce avatar 2.0 multi-angle edit
Playable on the hosted version
2.0 multi-angle AI edit
Extra camera angles layered over the delivery, showing detail and fit.
INDUSTRY NEWS

Exhibited at the Global Digital Trade Expo, showing AI presentation and multilingual avatars in cross-border trade.

INDUSTRY NEWS

Resident at the Yiwu Global Digital Trade Center, serving Chinese merchants producing multilingual content for global markets.

FULL CAPABILITY SET

Beyond those three, a dozen more essentials

AI script
Keywords in, full script out; rewrite existing copy into a viral structure with 30+ templates.
Text extraction
Pull the text out of video, audio, images or documents — from a link or a local file.
Smart cut
Stumbles and breaths detected and removed; text-based editing for people who don't edit.
AI erase
Remove text and watermarks from footage without a trace, leaving clean usable material.
Subtitles
Accurate recognition in one tap, with auto-generated titles and canvas backgrounds.
AI denoise
Strip background noise and enhance the voice — studio-ish audio at no extra cost.
Virtual background
Real-time matting without a green screen plus pro keying with one; 1,000+ backgrounds.
AR actors
Built-in 3D characters that share the frame and interact with the presenter.
Beauty & make-up
50+ presets built for talking-head shooting, plus filters and styled looks.
Sentence retake
Redo a line, pause anywhere to change person, outfit or background, join seamlessly.
Live prompting
Floating prompter plus Bluetooth remote; switch scripts mid-stream.
Training
Live classes and an on-demand academy that walk beginners through their first video.
USE CASES

Four kinds of people use it every day

UGC creation

UGC creators

Creators on Douyin, Kuaishou, WeChat Channels and Xiaohongshu use the single write-shoot-perform-edit flow to capture ideas as they come and build a personal brand.

E-commerce

E-commerce sellers

Photograph the goods and upload; the AI casts a model, sets the scene, writes the copy and cuts the video — selling clips at scale with no live shoot.

Education and training

Education & training

Trainers build courses with the prompter, virtual backgrounds, AR actors and automatic subtitles — or send an avatar on camera instead and go from script to multilingual course directly.

ADVERTISING

Small businesses, brands and agencies use avatars, AI scripting and smart editing to land creative ideas quickly, produce ad creative in volume, and generate multilingual versions automatically so a product reaches audiences worldwide.

WHAT USERS SAY

4.9 / 5 across app stores, 30,000+ ratings

Selected genuine reviews from the app stores, translated from Chinese.

★★★★★

"The teleprompter is extremely strong — it follows my speaking pace, so I never have to memorise lines and my delivery stays natural on a live stream. A few keywords and the AI writes the copy. The virtual background mattes in real time with no green screen; my stream instantly looks professional."

X不良帅X · vivo App Store · 2025-09-11
★★★★★

"With Shuode's AI teleprompter my live streams took off. It genuinely keeps up with how fast I talk, and the floating prompter means I can read straight from the script while streaming instead of racking my brain to memorise it."

User 269113514 · OPPO App Market · 2024-07-02
★★★★★

"The prompter suits me best. My eyes aren't good — I'm an older user — and the text scales up and down, and it adapts to how fast or slow I speak. Very thoughtful. Shooting with the AR actors is fun too."

遠囍 · Apple App Store · 2025-06-23
App Store Huawei AppGallery Xiaomi GetApps OPPO App Market vivo App Store
COMPANY

Hangzhou Gaishi Technology Co., Ltd.

Founded in 2017 and led by a former senior scientist at Google and computer-vision specialist together with the former CEO of the VR company Visbit, the company focuses on research and applied innovation in audio and video processing — so that ordinary people can express themselves through video easily, efficiently and cheaply.

🎓

A research team

The core team comes from Google, Microsoft and Amazon, alongside product and marketing talent from leading Chinese technology companies such as ByteDance and NetEase.

🏛️

One third hold advanced degrees

More than a third of the team hold master's or doctoral degrees from Columbia, Cornell, USC, HKUST and the Australian National University, among others.

🏅

Recognitions

Selected for Hangzhou's "Fledgling Eagle" programme in 2020; certified a National High-Tech Enterprise in 2023.

How the product got here

2021
Launched the in-house flagship Shuode Camera — a smart shooting app whose AI teleprompter follows the speaker's pace with industry-leading accuracy, becoming the default mobile prompter for Chinese creators.
2022 – 2025
Added virtual backgrounds, AR co-star actors, AI scriptwriting and rewriting, smart cutting and denoising — progressively covering the full path from shooting to edit.
August 2025
Shipped Photo-to-Speech, making portraits, animals and illustrations talk — the breakthrough into avatar-delivered video.
September 2025
Rebranded to Shuode AI and launched AI avatars, supporting both custom clones and a public avatar library.
25 September 2025
V6.0 shipped the world's first multilingual AI avatar — video generation fused with video translation, producing avatar videos in 27 languages in one pass.
Today
Shuode AI has grown from a smart camera app into an all-in-one AI platform for writing, shooting, performing and editing talking-head video, a category leader in China with more than 5 million active users.
PARTNERSHIPS

Bring us a real scenario and we'll build it in front of you

Whether it is content production at scale, multilingual selling for export markets, or wiring this capability into your own workflow — the fastest way to evaluate it is to hand us a script you are using today and see what Shuode AI makes of it in ten minutes.

shuodeai.com Back to the capabilities