AI Assistant

"Search has evolved from a tool for finding answers to a means of achieving outcomes. Traditional keyword-based search forced users to break their goals into fragments, often requiring multiple pages and query refinements. Chat-based AI improved this by offering synthesized responses, but still fell short of completing the user’s goal.

What if the system understood the end goal and helped achieve it directly? What if we designed for goal completion, not just information retrieval?

To answer these questions, we rethought the interaction model and layout—exploring adaptive, goal-based interfaces that anticipate user intent, support different interaction modes, and bring users closer to their goal with less friction."

RESEARCH

The Interaction Challenge

To inform the design of a more goal-oriented AI interface, we started by stepping back and asking two foundational questions:

Why do users search today?

Search isn't just about finding information—it's about moving toward an outcome. We analyzed behavioral patterns, ran user interviews, and mapped search intent across a range of everyday and professional scenarios.

We identified eight primary intent types that drive user search behavior:

RESEARCH

Key Insights

Most people aren’t searching for answers. They’re searching to accomplish something.
Traditional keyword search helped them begin the journey. Chat-based AI helped them understand the topic. But users are still left to assemble the final result themselves—be it a document, a comparison, or a decision.
Users don’t want to chat with an assistant. They want to collaborate with one.
This realization became the foundation for everything that followed—from the UI architecture to the interaction design patterns we developed. Our next step: designing a system that understands user intent and gets them closer to their goals with less prompting, less clutter, and more structure.

IDEATE

Different Modes of Interaction

Most people aren’t searching for answers. They’re searching to accomplish something.
Traditional keyword search helped them begin the journey. Chat-based AI helped them understand the topic. But users are still left to assemble the final result themselves—be it a document, a comparison, or a decision.