01 The road ahead,to be explored with users
Interviewer: What are the user habits and patterns when using the device?
Yang Sun: Which days of the week do you think user activity would be higher?
Interviewer: Weekends?
Yang Sun: Quite the opposite, and this also exceeded our expectations. We originally thought people would have more time and things to do on weekends, but the data shows that weekends are a low point, while weekdays are the peak.
People might really be using it as an AI copilot, not as a camera. The scenes in the videos, like petting cats, commuting to work, and working from home, are all from weekdays. There are certainly weekend uses, but I think people have more camera options on weekends. During the week, they use it as a copilot to check their daily life insights.
Interviewer: Which scenarios do you consider to be high-value use cases?
Yang Sun: The way users interact with AI is showing a completely new form. Many people are no longer just asking factual questions, but are starting to ask decision-oriented questions like, "Based on my state today, how should I adjust for tomorrow?"
There's one case that left a deep impression on me: a user was driving from Minnesota to a small town. Before setting off, he checked Google Weather, which showed average conditions. On the way, Looki identified the specific highway section he was on, warned him of an approaching snowstorm, and advised him to refuel at a suitable location before the storm hit. The prediction was completely accurate, and he later shared this experience on Discord.
The essence of this type of experience is that Looki has already developed a certain level of "contextual understanding" and can make inferences based on current environmental information.
Interviewer: Did you envision these scenarios before?
Yang Sun: I definitely thought about them, but it's no use just telling users about them. They need to discover it for themselves. Without experiencing that "aha" moment, they won't have such a deep feeling.
Interviewer: It feels like exploring a new map. You and your users are on the same boat, mapping a new continent together.
Yang Sun: Exactly. This is a brand-new scenario, a brand-new experience. It's like when the iPhone was first released; everyone thought the built-in apps were the ceiling, but even today, the iPhone's potential is still being discovered.
Interviewer: What do you think is the AI-native mindset?
Yang Sun: It's "less structure, more intelligence." You just need to set up the "field," rather than defining specific use cases or defining exactly what features to give the user.
For example, Notion set up a "field" by taking everyone's documents, notes, and other data, giving people sufficient but bounded freedom to see what chemical reactions would occur. Looki is the same. We bring life data together with AI to create a "field," rather than explicitly telling the user: "Sorry, my product is specifically designed for meeting minutes, and the AI feature I give you is to summarize meeting minutes." No, that's not the AI-native mindset.
It's the same as developing people. As the saying at ByteDance goes, "full context, not control." It's not about treating a person like a cog in a machine and telling them: you must execute this task.
02 Go slow,don't flip other people's tables
Interviewer: On the software side, what have been the key updates during this period?
Yang Sun: During this time, we've built a highly flexible "Inspiration" system.
It's essentially a low-perception, fast-iteration platform that is invisible to the user on the front end. Whenever we discover high-frequency, common needs in user-AI interactions, we can very quickly launch corresponding content in the "For You" section, such as modules for Daily Insights and emotion analysis.
We spent a lot of time in the early stages building this "field," and now we are mostly just quickly adding blocks on top of it. Once a module is established, the speed of combination and iteration becomes very fast, which is currently our small competitive advantage.
Interviewer: How did the comic mode come about?
Yang Sun: We had the idea for comics very early on, even before the launch. When new models come out, they greatly increase the success rate of some of the features we're testing. For example, comics were launched because we felt the time was right with the release of Nano Banana.
We have many more features lined up in the background. Life Data + AI—the potential is huge. The rest just depends on when the models are ready, and then we can launch them.
Interviewer: Does user feedback lead you to de-prioritize some features?
Yang Sun: Yes. We initially developed many features for the Lifelogging community. But because the product broke out of that niche right after launch, and it wasn't just lifeloggers buying and using it, the importance of features originally serving this group, like the "Memory" module, decreased. Now it just sits there, like a "cloud drive."
Interviewer: Looki's software design doesn't include many growth-oriented features. For example, "sharing" seems to have a low priority. The overall feel is very subtle. Was this intentional?
Yang Sun: We believe that growth must be driven by a signal. It's not something that's designed, but rather a clue that naturally emerges from real usage after the product is released.
For a startup, I'm always very cautious about growth—it's either a success or a failure. The most dangerous thing is to get stuck in the middle: no scale effect, yet unable to give up, easily copied, and most draining for the team.
So on this matter, we have to think very clearly about how to create network effects and scale effects. We will definitely do it, but it's still in the works.
Interviewer: I've had the product for about a week, and I told you: it's a product built for the silent user, and that really impressed me. Was this your original intention?
Yang Sun: Yes. Actually, it's not so "inward-facing" anymore. We originally thought of it as a 100% "inward-facing" thing. Although I'm an extrovert and occasionally post content from Looki, it's still mostly within my circle of friends. I never thought about publishing it very publicly.
But we've seen that many users, especially overseas users—who we thought valued privacy—don't seem to mind and are quite outward-facing. You can see in the community that they don't blur their family members' faces. There's already a component of being "outward-facing." This is why we believe there's huge potential for growth, social features, and community. Of course, these are still in development.
Interviewer: What is the typical user persona for Looki right now?
Yang Sun: Looki's early adopters are definitely AI early adopters, but it won't stay within the geek circle forever. It will gradually move towards a group of people who are not content with a mundane life and remain curious.
The growth path for new-generation AI hardware shouldn't replicate the smartphone—it doesn't compete for attention. Products like "Doubao Phone Assistant," while seemingly reasonable, are actually trying to do something that phones already do well—to flip someone else's table.
Today's AI carriers should enter the "slow value" domain where phones are not adept: emotions, introspection, and long-term companionship.
Interviewer: For this group of people, what is the most important value Looki can provide?
Yang Sun: What users truly love isn't the vlog or the comic grid itself, but the "surprise."
Looki breaks the interaction model people are used to: "I give a command—you give feedback." The user does nothing, and the next day receives unexpected, interesting content. This experience of being proactively "served" content is the core of its appeal.
The content format can change, but the logic is clear—continuously and naturally create surprises for the user, and then intersperse high-value information within them. This is how stickiness is created.
