How an AI Short Made With Veo 3 Hit 80 Million Views in 5 Weeks
A case worth studying recently came out of the online education industry.
Yanadoo, one of South Korea's largest online English-learning platforms, published a video made with Google's AI video generation model Veo 3 — and it racked up 80 million views in just 5 weeks.
The moment Yanadoo released the video, education brands and competitors rushed to produce similar AI videos — and today, clips in that same style have flooded the short-form feed.
What made Yanadoo's content blow up?
✔ Sharp planning and storytelling with a twist certainly helped,
✔ but above all, the scene of a Black man and an elderly grandmother talking to each other was rendered with remarkable precision and realism.
Yanadoo's CEO and a member of its AI team later discussed the campaign in an interview on Sebasi, a well-known Korean talk show —
where they explained that they built the content using the AI tool Veo 3.
Below, we'll take a closer look at Veo 3
and how you can put it to work.
What Is Google Veo 3?
Veo is the name of Google DeepMind's family of text-to-video AI models — its latest generation of video AI, a clear step up in capability from earlier video generation models.
Veo 3 is the newest version in the family, released around May 2025.
Its defining feature over earlier versions is the ability to generate video with synchronized audio.
▶ Key features of Veo 3
Earlier Veo models generated video only, with sound often bolted on separately —
but Veo 3 is designed to generate video together with dialogue, sound effects, and background audio.
That said, much of it is still offered in preview form, and it isn't yet fully open for unrestricted use.
Veo 3's output is also typically limited to around 8 seconds per clip, so for now it's used mainly for short clips rather than longer videos.
Where You Can Make AI Shorts With Veo 3
There are several tools that give you access to Veo 3.
Google's Gemini
Flow
DeeVid AI
Filmora
All of these either offer the Veo 3 model directly or use it under the hood.
▶ Gemini
Gemini is Google's multimodal generative AI platform, capable of understanding and generating text, images, audio, and video.
In other words, it's not just a chatbot — it's a unified model that can read text, interpret images, and generate or summarize video.
▶ Flow
Flow is Google's AI video creation and editing tool.
It integrates Google's generative models — Veo 3, Gemini, and more — into a tool for generating and editing storytelling-driven video.
If your goal is to turn text into short-form video, Flow is the better choice of the two.
It has Veo 3 built in, a video-first UI, and a workflow optimized for social short-form formats.
▶ DeeVid AI and Filmora
Beyond Google's own platforms, you can also use Veo through DeeVid AI and Filmora. The appeal of these third-party platforms is that they let you do more with the model.
For example,
DeeVid combines multiple models to automatically handle video generation, audio, and captions in a single pass,
while Filmora pairs generation with a full editing workflow,
so you can combine it with traditional video editing for far more flexibility.
How to Make AI Shorts With Veo 3 — and How to Write Prompts
More important than which platform you pick is
writing the right instructions — the prompt — so the model produces exactly the video you have in mind.
The prompt is what determines the quality of the output:
it controls every element of the video — the scene, characters, setting, camera angle, movement, and expressions. Learning to write good prompts is essential.
4 essential elements every prompt should specify
What to show
Example: A Korean lawyer in his mid-30s stands beside his desk in a bright office illuminated by fluorescent lights.
What moves, and how
Example: He holds a coffee cup in one hand while facing slightly toward his computer.Camera angle, lens, style, and lighting
Example: The camera captures a side view of him in a well-lit office environment.Time of day and mood
Example: It is around 3 p.m., and his face looks slightly tired, reflecting the calm afternoon atmosphere.
Things to keep in mind when writing prompts
Veo performs noticeably better with prompts written in English — if English isn't your first language, use ChatGPT to translate
Avoid vague words like "pretty" or "nice"
Be as specific as possible about appearance, expressions, tone of voice, and gestures
Keep it simple, concise, and unambiguous
Yanadoo's 80-million-view result with Veo 3 is more than a viral one-off —
it's proof that AI short-form video production can deliver real business results.
The old assumption was that "AI video is experimental and not quite there yet" —
but it has now reached a level that's genuinely usable across marketing, education, and entertainment.
Going forward, the creators and teams who learn AI tools fastest — and use them most effectively — will be the ones whose content wins.
For brands and creators alike, the ideal strategy is to build up production skills with AI short-form tools like Veo 3
and move quickly toward a system for producing content at scale.