The Breath Guide


The Breath Guide

Spire's most popular feature. Using visuals and interactivity to practice calm breathing.


The problem:

Our users depend on their Spire Stone to manage their anxiety and tension. When users begin to have an anxiety attack or intense tension, they play an audio meditation to bring them to calm. However, audio tracks do not adapt to the user’s condition and may end even if the user has not reached calm. 



Create a guided breathing exercise that (1) is aware and adaptive to the user’s state and (2) lasts as long as necessary, only ending when the user has reached calm.



I worked on all aspects of the project including product spec, user testing, iterations, final visual design, and engineering handoff. I worked close together with our industrial designer and firmware engineer who were passionate about the project, as well as with our VP of Product on user testing.



VP Product
Chief Science Officer
iOS and Firmware Engineers
Contract Visual Designer


What I Learned

Designing the Breath Guide exposed me to the hidden engineering complications involved in creating a seamless experience for the user. I worked closely with engineering to discover edge cases, set parameters for the designs, and define rules for how the product’s interactivity should work.

Working closely with our Chief Science Officer, I learned how to design a product that incorporated and reinforced best practices recommended by the scientific community.


Background: the Spire Stone

Our first product, Spire Stone, was a hit amongst users who had tension and anxiety.

The Spire Stone knows in real-time whether its user is tense, neutral, or calm, based on their respiratory patterns.

Spire Stone would discern when users were tense and then alert them. Users’ favorite feature was the Spire Stone vibrating when the user’s breathing became tense.


The Problem: Guided Meditations

Our user research indicated that after their Spire Stone vibrated, users would often use an audio meditation exercise to reach calm.

Audio meditation exercises have several shortcomings, most notably: 

  1. Since they are pre-recorded, they have a defined end. Even if the user is still having an anxiety attack or experiencing serious stress, the exercise is over.
  2. Many audio meditations have various stages in the exercise. If a user’s anxiety or tension is not improving, however, it does not make sense for them to proceed from one stage of an exercise to the next. 

Both of these problems cause additional stress to the user who is now worried about not keeping up with the meditation track.


Our Goal

How can we leverage the Spire Stone’s realtime monitoring to improve the experience of a guided meditation?

Using Spire’s realtime monitoring, we found an opportunity to create an exercise that would know how a user was feeling and use that information to provide tailored guidance, continue until the user reaches calm, and end at the appropriate time.



What should the most basic version of this product look like?

We quickly discovered that adaptive audio would be the most complicating factor. So we started with testing visual-and-text-only exercises to validate whether the idea of an intelligent meditation exercise had merit.

We took inspiration from existing apps with calming visuals and began testing and iterating on:

  1. Visual styles
  2. Text guidance. What should it say?
  3. Stages of the exercise. When should it say it? 
  4. Rules for success. When should the exercise end?


Saying we'd build an "intelligent" guided exercise was easy. Making decisions on parameters for how the exercise starts, ends, and progresses was quite complicated.

For example: what if the user’s anxiety is so severe that the exercise is not helping them make any progress, even after several minutes of practice?

Should the exercise adapt it’s instruction? If a user notices the adapted instruction and realizes they are not progressing, will that cause them more stress? Is there a certain point at which we have a responsibility to recommend the user seek more serious help? 

Another challenge: in our testing we quickly discovered that users would be interested in using this adaptive exercise even if they were not experiencing tension at the moment. 

If a user begins the exercise with no tension, how should we determine when the exercise should end? Percentage change in respiratory rate is no longer a sufficient metric, because a tense user beginning the exercise with 18 breaths per minute will realistically reach 9 breaths per minute. But a calm user beginning the exercise with 8 breaths per minute may not reach 4 breaths per minute. 

Solving these challenges required working closely with our Chief Science Officer and engineering team. Our Chief Science Officer provided guidance on best practices recommended by the scientific community for addressing tension and anxiety. Engineering helped us define what was possible and helped us think through the various mathematical challenges involved in creating the intelligence to power the exercise.


Final Designs