Background
I was initially hired on at Capacity to build out the personality of their AI-powered virtual assistant. While stand-up comedy and screenwriting had afforded me a solid foundation of character development experience, my knowledge of virtual assistants was limited to the occasional, “Hey Google, could you, um, Google…”
I knew that if I was going to create a virtual assistant, I’d need to, well, become one. And to become one, I’d need to get to know some virtual assistants.
So, I interviewed them.
I developed a 25-question survey designed to measure the personality—origin story, identity cohesion, response variants, etc.—of a dozen of the leading virtual assistants, in addition to the client's direct competitors.
I posed each query until failure, which was defined as either three consecutive repeat responses, two complete cycles through any associated response variants, or no response at all.
Below are some of my findings, followed by a selection of personality exchanges I subsequently designed.
Role
I performed the research, synthesized the data, created the data visualization, and designed the virtual assistant's personality.
Deliverables
Deliverables
• Virtual Assistant Personality
• UX Research
• Market Research
Skills
• UX Research
• Market Research
Skills
• UX Research
• Conversational UX Design
• Personality Design
• Conversational UX Design
• Personality Design
Synthesis Terminology
• Query: Any prompt—whether a question or a statement—used to elicit a response from a virtual assistant.
• Response Variants: The multitude of ways in which a virtual assistant can respond to the same query. For example, the query, "Hi," could evoke response variants such as, "Hello," "Hey there," "Yo," etc.
• Response Rate: The frequency at which a virtual assistant returned a logically consistent response to a given query.
• Response Variants: The multitude of ways in which a virtual assistant can respond to the same query. For example, the query, "Hi," could evoke response variants such as, "Hello," "Hey there," "Yo," etc.
• Response Rate: The frequency at which a virtual assistant returned a logically consistent response to a given query.
Virtual Assistant Personality
Armed with this data, as well as a newfound, somewhat eerie connection with virtual assistants, I built out Capacity’s personality.
In my opinion, the essence of the workplace virtual assistant—proudly mechanical, infatuated with productivity—really shines through in the questions, “Can you sing?” and, “What are you passionate about?”
Keep in mind that the following selection of exchanges highlights the fun part of conversation design. The much more time-consuming and challenging conversational UI efforts are akin to a game of chess, in which you’re constantly attempting to think 7 dialogic steps ahead of the user.
Unlike chess, however, there are few agreed upon rules, and most victories go largely unnoticed.