VOCHI: what designing an AI video app that became App of the Day in 126 countries (and joined Pinterest) taught me
Alla Kuchynskaya · Senior Product Designer
· 7 min read
Short answer: VOCHI was an AI video editing app from Minsk that applied effects to moving objects in a video. I was its UI/UX designer in 2019–2020: early concepts, wireframes, prototypes, the first design system, usability testing and the iOS and Android UI. It became App of the Day in 126 countries and was acquired by Pinterest in 2021. My main takeaway: in an AI app, the effect is the product, and the interface exists to get people to it fast.
What was VOCHI?
VOCHI was founded in 2019 in Belarus by Ilya Lesun, Anna Buglakova, Vasily Kasnitsky and Sergei Malyutin, and incubated by Bulba Ventures, according to TechCrunch. Its core was a computer vision model that segmented objects in a video, so an effect could be applied to one person or one object instead of the whole frame. Typical effects were motion trails, glitches, clones and lighting changes.
In November 2019, iDownloadBlog picked it as an app to try and described it as offering "unique AI-based technology effects never seen before." In its 2020 seed round coverage, TechCrunch wrote that users could see effects in real time before downloading the high-quality video.
The public numbers after that:
- By mid-2021, around 500,000 monthly active users and roughly 20,000 paid subscribers (TechCrunch).
- A $4M+ annual run rate, with about 60% of effects and filters free (AIN).
- In December 2021, Pinterest acquired VOCHI. The 40-person team joined Pinterest and the app stayed available as a standalone product (TechCrunch).
What did I do on VOCHI?
I joined in May 2019 and stayed until February 2020, during early product development. My work covered concepts, wireframes and prototypes, the initial design system (built to scale as effects were added), usability testing, and leading UI design for iOS and Android around the AI video effects. VOCHI became App of the Day on the App Store in 126 countries.
To be precise about credit: the acquisition happened almost two years after I left, and it belongs to the founders and the team who kept building. What I can speak to is the early stage, when the product first took shape on screens. That stage shaped how I think about AI products today.
How does an app get featured as App of the Day?
I can't tell you how Apple's editors made their decision about VOCHI, and I won't pretend to. What is public is Apple's own list of featuring considerations: user experience, UI design, innovation, uniqueness, accessibility, localization and the product page. Teams can nominate an app through App Store Connect, with at least two weeks' notice and ideally up to three months.
Reading that list now, I notice that "innovation" and "uniqueness" are things the technology gives you, while "UI design" and "user experience" decide whether anyone can see that innovation in the first few seconds. A new AI effect hidden behind three setup screens is not unique to a reviewer. It's just slow.
Six lessons for teams building AI consumer apps today
These are my lessons, from VOCHI and from the AI products I've designed since.
1. The effect is the product, so the UI has to step back
In a video app, the content is the screen. Every panel, label and toolbar competes with the thing people came for. I'd still start any AI media app by asking: what is the smallest set of controls that lets someone get a result they want to share? Everything else can live one tap deeper.
2. Design for the "wow" in the first seconds
People decide whether an AI app is magic or a gimmick almost immediately. So the metric I'd watch is time to first result: how long from opening the app to seeing your own video transformed. Good defaults matter more than many options. A first effect that is already tuned to look good beats a slider at 50%.
3. Make heavy AI feel instant
Segmentation and rendering are expensive. The user shouldn't feel that. VOCHI's public positioning was a real-time preview first, then the high-quality export. The general principle: show something useful right away, even at lower quality, keep the canvas responsive, and save the waiting for the moment the person has already committed (export, save, share). If something must take time, say honestly what is happening.
4. Start the design system early, even if it's small
AI apps grow by adding capabilities. Each new effect needs a thumbnail, a name, a preview state, adjustable parameters, a loading state, an error state, and often free and premium variants (remember that roughly 60% of VOCHI's effects were free). If "effect" isn't a component from the start, the tenth one will look like it came from a different app. An early system is less about pretty buttons and more about making growth cheap.
5. Test on real content, not on demo clips
A demo has to work once, with the author watching. A product has to work on whatever people actually shoot: dark rooms, shaky hands, three people in the frame, a cat. AI behaves differently on messy input, and the interface needs states for when the model is unsure. In usability testing I'd ask participants to use their own camera roll, because a polished sample video tests the model's best day, not the product.
6. Treat the free/paid line as a design decision
With a subscription model, where the paywall appears is part of the experience. My rule: let people reach the "wow" before asking for money, and make the premium effects visible and tempting rather than locked and grey. A paywall that comes before value feels like a toll. One that comes after feels like an offer.
A quick checklist for designing an AI video app in 2026
| Question | Why it matters |
|---|---|
| How many taps from launch to the first transformed video? | This is your real onboarding |
| What does the user see while the model works? | Waiting without feedback reads as broken |
| What happens when the model fails or is unsure? | Real footage is messy |
| Is an "effect" a reusable component with all its states? | You will ship dozens of them |
| Where exactly does the paywall appear? | Before value it costs you trust |
| Was it tested on participants' own videos? | Demo clips hide the hard cases |
FAQ
What was the VOCHI app?
VOCHI was an AI-powered video editing app from Belarus that used computer vision to apply effects to individual moving objects in a video, such as motion trails, glitches and clones. It ran on iOS and Android and had about 500,000 monthly users by mid-2021, according to TechCrunch.
Why did Pinterest acquire VOCHI?
According to TechCrunch, the deal was part of Pinterest investing in creator tools. Pinterest's Bin Liu said the team would help "bring more video creation tools and quality content" to Pinners. The 40-person team joined Pinterest in December 2021.
How do I get my app featured as App of the Day?
Submit a featuring nomination in App Store Connect, ideally up to three months ahead and no later than two weeks before. Apple's published criteria include user experience, UI design, innovation, uniqueness, accessibility, localization and a strong product page. In my view, the innovation only counts if a reviewer can feel it in the first minute.
How do you design an AI video editing app?
Put the content first and keep controls minimal, optimize for time to first result, show a fast preview while heavy processing happens in the background, build effects as reusable components from day one, and test on real user footage. Then decide carefully where free ends and paid begins.
Alla Kuchynskaya is a Senior Product Designer and AI Product Designer & Builder with 12+ years in startups. She worked as UI/UX designer on VOCHI (App of the Day in 126 countries, later acquired by Pinterest) and designed the chat experience for the Flo women's health app. She also runs the YouTube channel Ala pro AI, about AI through a product designer's eyes. Based in Bali, she works part-time, full-time or per project: alla.studio, Telegram.
About the author
Alla Kuchynskaya · Senior Product Designer
alla.studio · LinkedIn · Telegram · Resume