the backstory

starting steps


Voice UI Experience Design | Class Project

"I wish my Keurig wasn't such an enabler. It's too easy to drink too much caffeine."

This was a comment from a coffee-loving friend who was frustrated that her Keurig always provided quick and easy coffee and never told her no. Which got me thinking...

What if working machines had a sense of judgement? What if they could gauge whether performing the requested task is or isn't in the user’s best interests, based on what the user says or the user’s behavior?

The average American consumes over 3 cups of coffee a day - and over consumption can lead to anxiety, tremors, restlessness, and so much more. This project explores how a Keurig might regulate caffeine intake by reacting to emotional and health cues from the user.


“I need a cup of coffee.”
“Coffee, tall, extra hot.”
“Give me a coffee.”
“One coffee.”
“A tall coffee.”
“A cup of joe!”
“Can I get a coffee?”
“I need caffeine.”
“I’d love a coffee please.”


To consume coffee safely

  • Because the Keurig is a personal coffee machine located at home, it is very familiar with the user’s coffee drinking patterns and preferences such as size and brand.

  • Through the first few weeks of use, the Keurig notices if the user strongly prefers a particular order and will inquire if the user would like to make this their usual.

  • The Keurig also has a decaf option that the user sets during onboarding.

  • This particular Keurig model is marketed as a machine that helps regulate caffeine intake - so the target user is someone who wants help in preventing bad caffeine habits.

How might we help users consume coffee safely in a way that feels guiding rather than constricting?

conversational modeling - happy flow

User: “I’d like a coffee.”

Keurig: “Okay. Would you like your usual order?” (Unspoken variables: Coffee Type and Size)

User: “No, I’d like the smaller size today.”

Keurig: “Great - A small Donut Shop Medium Roast. Do I have that right?”

User: “Yep!”

Keurig: “Coming right up.”

mapping the flow

The flow snippet below maps the interaction between the Keurig and a user who has already onboarded and has been using this Keurig for a while.

Click here to download the full document.


video - accounting for rudeness

Though there were many paths of interaction to prototype, I mocked up the path of user "rudeness/irritability," which triggers the Keurig to withhold caffeine.