Daniel K. Adam
Product Designer

Conversational over chatbot

Initially, we explored different interactions, including a traditional chatbot UI.
However, chatbot interfaces often felt too impersonal.
Instead, we opted for a conversational UI approach that prioritized a natural flow.
Users speak freely without being confined to an input field.

Interruptions: Keeping Conversations Dynamic

Users needed the ability to interrupt the AI naturally, just like in human conversations. This required real-time processing to handle interruptions smoothly and visual and audio cues to indicate when the AI was actively listening or responding.

No Push-to-Talk: Reducing Interaction Barriers

A major design decision was eliminating the need for a push-to-talk button. Instead, the Agent would detect when a user started speaking and respond accordingly. To avoid accidental activations through sensitivity tuning.

Voice-First: Prioritizing the Speech Experience

Since voice was the primary mode of interaction, the UI needed to support a voice-first experience. The interface minimized reliance on text-based inputs while still offering a fallback to chat.

Building Trust with a Live Preview

I designed an Agent creation flow that guided users step by step through the setup process. On the right-hand side, users could see a live preview of their Agent, providing immediate feedback on how their choices impacted the appearance and behavior of the Agent.

Embedding flow

I designed an embedding flow that lets users choose between a section div or a floating widget, preview how it will look on their website, and customize branding, position, and visibility to fit their needs. This made integration feel trusthworthy.