AVA + Soul Machines Beta
How often have you sent someone a text message and had that message be understood in a way completely different than the one in which you intended? There are many visual and auditory cues we receive from a face-to-face or phone conversation that simply aren’t there via text. The same holds true for chatbots. What if chatbots were capable of voice and video chat and could recognize and respond to a customer’s emotional cues?
In an effort to explore how voice and video capabilities might enhance the chatbot experience, we partnered with New Zealand-based company, Soul Machines to develop a digital avatar for AVA. We piloted our emotionally enhanced digital AVA experience as an alternative to certain complex, multi-step troubleshooting articles.
My Role UX Lead
Activities Wireframing, documentation, interaction design, visual design, UX validation, and additional UX support as needed.
Project Team UX research, business analytics, conversation engineers, application developers, Q/A testers, and product management.
I needed to answer the following questions and address the following needs:
- Solution that can work with video chat, voice or text-only modalities
- How do we tell the user which modalities are on/off?
- How can the user toggle modalities? What should each modality experience look like? What should the transition experiences look like?
- Should the user be able to see themselves?
- How does the AVA avatar “come to life”? (Expand)
- How big should that UI element be
- How can we mitigate the impact on the chat interface?
- How can we deliver/present supplemental information to the customer? (e.g. text blocks, visual aids, etc)
In response to the above requests, I spent time reviewing Soul Machines documentation and best practices. After my initial assessment, I began to explore solutions. I decided it made the most sense to approach the UI as if it were a video conferencing solution with the same types of controls.
Vivamus ipsum velit, ullamcorper quis nibh, molestie tempus sapien. Mauris ultrices, felis ut eleifend auctor, leo felis vehicula quam, ut accumsan augue nunc at nisl.