Introduction of Assistance in Conversational Experience Product Design • Interaction Design

backgroundIn the Crux Copilot beta version, responses were generated for users based on certain assumptions, leading to a subpar experience with accuracy issues occurring approximately 8 out of 10 times. To address this challenge, we opted for the explore and exploit strategy. This strategy involves initially understanding the user's preferences and then making assumptions based on the collected data.

design objectiveThe goal is to enable the system to receive assistance from the user whenever necessary. To achieve this, we introduced the "assistance feature." This feature allows users to make specific selections, enabling the system to generate accurate responses based on the provided input.

An adaptive morphing component that seamlessly transitions between the ask bar, assistance, and loading states, dynamically responding to the user's journey toward receiving a response.

We wanted to let the user toggle between collapsed and expanded state depending on their preference. Longer inputs meant it would be easier for the user to have an open view while shorter inputs wouldn’t require an expanded view.

The component was designed to be scalable, with all elements being optional. This makes it suitable for use by external teams or clients to build upon.

Focused case studies from Crux Copilot

generative ai Crux Copilot Introduction
under the hoodBringing transparency in responses - How did we arrive at this?
error handlingDesigning to Seamlessly Guide Users Through Technical Hiccups

let’s create together!  
If you’re looking for your next Designer, need a Collaborator or just want to have a conversation, I’d love to meet you!

Let’s Talk! 

contact mail