Designing Better Feedback for AI Talker

AI Talker was an AI-based conversational service where I worked on frontend development and mobile interaction improvements.

One issue was that users could not always tell what the system was doing. When a response took time or voice functionality changed state, the lack of visible feedback could make the service appear unresponsive.

I worked on clearer response and voice states so users could better understand whether the system was processing, waiting, or ready for the next interaction. I also worked on the mobile chat experience and the migration to Next.js.

The project reminded me that feedback becomes especially important in AI products. Traditional interfaces often respond immediately to an action, while AI interactions can include unpredictable waiting time and multiple system states.

Making those states visible does not change the underlying model, but it changes how users interpret the experience. Sometimes a small piece of feedback is what makes the difference between “the system is working” and “something is broken.”