🔍 Discover
When I joined this initiative in summer of 2024, the team was hard at work building a solution that was considered a strong MVP. Some initial usability testing had been completed using Figma prototypes, but had not yet tested how the Ambient model might hold up in a real world visit, or how customers would respond to the actual output of the LLM. With the expected launch of the feature 4 months away, we needed rapid validation and iteration cycles to feel confident in our design decisions.
We started the process with site visits to 3 customers, where we used the current Dev build (on a secure server that was refreshed daily) to shadow and summarize provider/patient visits. This enabled us not only to witness how a provider might use this solution - mobile phone and all - during their workday, but it gave us a unique opportunity to have providers react to an experience that was as close to their daily work as we could get. After each site visit, detailed notes were shared with the team, who eagerly used them to identify needed changes to their workflows.
After the visits, I created a robust presentation with some of the core findings from the visits, which provided foundational understanding for the team to draw from as they continued iterating on the solution. This included workflow diagrams that illustrated how patients and providers moved through their visits, and some "aha" insights that inspired shifts in the design approach.
📖 Define
After the site visits, I worked with one of the team’s designers to create a detailed, complex prototype in Figma that mimicked the full workflow from capturing on Mobile to seeing the summary and finalizing the note in Desktop (where providers are most likely to close their notes). To make this happen, I:
- Worked with our Patient Safety team and our Physician Solution Consultants to capture realistic patient/provider visits in our three starting specialties: Ortho, Women’s Health, and Urgent Care.
- Ran those recordings through our Ambient solution to generate summaries that could be pulled into the prototype.
- Created variables in Figma that would allow us to change content in the prototype depending on the type of visit.
- Set up the test protocol and prototype to go down a distinct path depending on the type of visit most relevant to the participant.
Using this prototype, we did two rounds of testing to refine the experience. The first, done over screen share, allowed us to iterate our prototype every week as we learned from users over four weeks. The second round, with our “finalized” prototype and some new potential enhancements, we tested at Thrive, athenahealth’s annual user conference, where the Ambient Notes feature was unveiled to a thrilled audience.
📐 Align
A project of this significance, which was already well underway when I joined the team, required a nuanced approach that didn’t sacrifice our Mobile team’s propensity to learn and iterate in rapid cycles. Beginning with a review of the previous research for this project and the site visits gave me an opportunity to get some foundational insights for the team quickly, and put those insights in context of users’ actual workflow. For example, we learned:
- that users of these types of tools think of them as a “second brain,” which makes it extra scary if the system fails, or if it looks like information is lost.
- that the rapid pace of visits means that most users will expect to send the summary to their note immediately after the visit is over and move on to the next visit.
- that users might also use the ambient tool to capture “prep” before the visit, such as chart review or reviewing the previous visit with the patient.
These insights helped the team continually refine the workflow over the four months leading up to launch. It also helped them envision how the full mobile-to-desktop workflow should behave.
