Tenshow Innovation builds AI-powered products for learning, communication and global users. Our four products look different from each other: an app for AI audio companions (Boilon), an AI teaching assistant (StudyOS), a spoken-English app (入戏英语) and an AI meme translator (MemeWingman). They share a way of working. These are the principles behind it.
1. Start from a real job
Every product begins with something a person is trying to get done, not with a model or a feature list:
- “I do not get this meme, and I need to reply.” leads to MemeWingman.
- “I want to speak English in real situations, not just understand it.” leads to 入戏英语.
- “I want to spend less time preparing and more time teaching.” leads to StudyOS.
- “I want one place to set up and manage my AI audio device.” leads to Boilon.
A good test is whether we can state the job in one sentence.
2. Design for many languages from day one
Global products need more than translated buttons. Language shapes content, tone, search, input and even what counts as a correct answer. MemeWingman explains memes in five languages while keeping each meme’s name in its original language, because the name is how people find and recognize it. 入戏英语 pairs English dialogue with Chinese explanations and cultural notes, so learners understand both the words and the situation. And this website exists in Simplified Chinese, English, Traditional Chinese, Japanese and Korean, because it should speak to the people who use our products.
3. Keep personal data close to the user
Where a feature allows, personal data stays on the user’s device. MemeWingman stores friend profiles, saved memes and chat history on the device. 入戏英语 keeps videos imported on mobile on the device by default and uses only the audio to generate subtitles. We still send what a feature genuinely needs: the messages you ask MemeWingman about go to the AI model so it can answer, and we say so in plain language. The principle is simple: collect and move as little as the experience requires.
4. Iterate on evidence
AI products can look finished in a demo and fall apart in daily use. For each product we pick a small number of key user paths, measure them, listen to feedback and ship small changes often. Metrics tell us where to look; users tell us why. We keep what works and drop what does not.
What this means in practice
- Products stay focused: each one does a few things well.
- New languages and regions are planned, not bolted on.
- Privacy statements describe what actually happens.
- Improvements are frequent, small and measured.
Talk to us
If you are working on something similar, or would like to explore a partnership, we would like to hear from you. Get in touch.