What are virtual bots currently like and what is their future potential?
We’re surrounded by bots and AI-infused virtual assistants in our everyday lives, or at least if you choose to be. For me, a typical day involves the phrase, “Echo, good morning” as soon as I wake up (preferring the more gender-neutral Alexa presence). This turns on my lamp, plays news highlights, and then seamlessly transitions into a radio station of my choice. As I get ready for the day I find myself constantly inquiring, “Echo, what time is it?” to find out I’m running behind. Later that day, if I have a question for my husband I say, “Hey Siri, text Gaines: Do you know where the phone charger is?” And in the evening while cooking dinner I yell from the kitchen to the Alexa device sitting in my living room, “Echo, play music!” and minutes later, “Echo, add carrots to my shopping list!” Ending the night with a simple, “Echo, wrap it up.” resulting in all of the houselights turning off.” I clearly live in the future, and it’s pretty dope.
As you can see, my life has been drastically simplified and enhanced by the presence of bots as virtual assistants. Now, on to the good stuff…
While I was a designer in IBM’s Social Collaboration organization back in 2014, I had the responsibility to understand the potential for our products (like email, chat, file-sharing, and e-meetings) to meaningfully leverage bots. So, as a starting point, I asked myself, “Laura, what are bots currently like and what is their future potential?” Unfortunately, the answer didn’t come spilling out of a device like I’m currently used to, but I built a framework that breaks it all down with some notes on what we feel confident we know, and open questions we still have.
Let me know what you think and if anything has changed in the last 6 years.
a. “I’m your virtual team member, listening and surfacing what you need.”
Knowns
• always listening
• high confidence suggestions pushed to the user in real-time
Unknowns
• concerned about the level of disruption
• what is the UI pattern? (ex: highlight, underline, sidebar…)
• how consistently does this need to be in the UI
b. “I wait for you to ask me to do something.”
Knowns
• user can pull information from the system
• includes invoking the system with text or toggling on a ‘mode’
Unknowns
• concerned about our findability
a. “I talk to your whole team at the same time.”
Knowns
• it needs to be clear when a team member is interacting with a shared auto-generated suggestion to avoid redundancy of efforts
Unknowns
• can a team model act differently than a user’s personal model, resulting in a system that doesn’t behave as expected
b. “I only talk to you.”
Knowns:
• best use case for voice to voice machine to human interactions
Unknowns:
• how does the user mentor the system to build their personal model
a. “I am your only bot and you call me _______.”
Knowns
• if we give the system a name, it would have to make sense from anywhere in the product
• aiming for personified, but not trying to be human
• this could act as the single-entry point to a multitude of capabilities on the back-end
Unknowns
• how the system gets represented consistently across multiple channels of communication
• how to disambiguate the user’s intent when back-end capabilities overlap in functionality
b. “I am one of many bots you can choose to talk to.”
Knowns
• usually results in bots being given names that relate to their function (ex: meetings bot, files bot)
Unknowns
• concerned about introducing a new “expertise locator” problem
c. “I don’t need a name, I’m just a better version of the system you have today.”
Knowns
• integrates cognitive solutions into the fabric of the UI
• replaces older/inefficient workflows, instead of just adding a natural language wrapper on existing functionality
Unknowns
• how to have 2-way communication with an unnamed system, along non-chat channels of communication
a. “You can find me on what was formerly known as the search bar, but now with enhanced functionality.”
Knowns:
• extends search into a 1:1 conversational format that is globally available
• typeahead and autosuggestions can help the user understand the system’s boundaries
Unknowns:
• how can 1:1 dialogues benefit or translate into a team context
b. “You talk to me using tools you usually talk to people with (like chat, email, or @ mentions.”
Knowns:
• user should be able to communicate with the system how they communicate with people, synchronously or asynchronously
Unknowns:
• how are the communication methods with the system obvious, and the boundaries of its assistance evident
c. “I am separate from your current tools and can be called from anywhere.”
Knowns:
• most prevalent use case would be invoking commands from the OS of your device, without having to go into a specific app
Unknowns:
• how this would co-exist with existing intelligent personal assistants, like Siri, Cortana, or Alexa
a. “I use my voice to talk back to you.
Knowns:
• best for simple requests
Unknowns:
• voice and personality
• can the user choose their voice
“I send you a text response.”
Knowns:
• for moderate requests
• potential for suggested actions in the text that users can opt-into
Unknowns:
• tone and personality
• how appropriate are emojis
c. “I surface an interactive card or micro-app.”
Knowns:
• for complex requests
• content should be pre-populated as much as possible
Unknowns:
• standard interaction patterns internally/externally
This “menu of options” as I like to call it really allows you to be purposeful about the design decisions you make building your bot experience. The last thing you want to do is assume every use case has to fit into the same delivery mechanism—missing out on more high-value opportunities. For our product, we wanted to leverage multiple bot-combinations to ensure our end-user were getting the most seamless and intuitive experience relative to their specific goal and touchpoint.
What else can I say? This stuff is super fascinating and it’s always satisfying to see my frameworks stand the test of time. 🥳
Back to the grind…
