From Chatbot to Conversation: Designing Ally’s First Interaction

At KH Creative, we think of design as less about pixels on a screen and more about how people feel when they interact with technology. With Alley Cat AI Works, that meant asking a simple question with complicated implications: what does it mean to trust an AI assistant?

Ally began as a straightforward website chatbot—an entry point for questions and answers. But like any good design story, the process revealed that we weren’t just building a widget. We were designing a relationship.

Listening First

We started with stakeholder and user interviews. The takeaways were consistent:

• People wanted clarity at onboarding (tutorials, expectations, a simplified flow).

• They needed trust (activity monitoring, transparency, visible “guardrails”).

• And they craved variety (light/dark modes, UI theming, and yes—voice).

Those sticky-note walls turned into our north star: Ally had to be simple, explainable, and flexible.

Learning from Patterns

Before sketching, I researched chatbot design patterns—from Otter.ai’s transcription flow to Replika’s avatar-driven conversations. A few things stood out:

• Clean, uncluttered interfaces build confidence.

• A conversational tone matters just as much as the visuals.

• The fastest way to break trust is when the bot doesn’t really listen.

That research gave us the foundation for Ally 1.0.

Ally 1.0: Getting Started

Our first design was intentionally light. A simple UI embedded in the site with the prompt: “Can’t find it? Ask Ally.” It was functional, easy to use, and aligned with the project’s initial scope.

But testing revealed something bigger: users didn’t just want to type to Ally. They wanted to talk. Voice wasn’t a feature request—it was an expectation.

The Pivot: Designing for Voice

This insight pushed us back to ideation. Ally 2.0 is being reimagined not as a chatbot, but as a voice-first assistant with multi-modal support. That means:

• Designing voice flows that feel natural, with room for interruptions and corrections.

• Offering UI flexibility—light/dark modes and brand-friendly theming.

• Keeping education at heart—Ally doesn’t just provide answers, it explains them.

This pivot reframed Ally as something larger: a conversation layer between humans and AI.

What’s Next

The work continues. We’re prototyping voice interactions, refining safety guardrails, and testing Ally across contexts.

Because designing AI isn’t just about interaction—it’s about creating experiences people trust, want to use, and can learn from.