Background
Company: Cisco
Name of Product: Beyond the GUI
Type: Solo/Team
Role: Researcher and Product Designer
Date: 2023
Let's Begin
This research served three purposes—the first was to learn how experts hear about new applications. Secondly, understanding the role of CLI, GitOps, and GUI—lastly, AI roles in the expert work environment.
An outside service provides the participants, candidates, and Moderator. We furnished the interview script. We reviewed it with the Moderator to see if he had any questions. The experts' profiles were reviewed, and we selected the appropriate ones. The interview took place over five workings. After each interview, the agency sent the transcript within twenty-four hours.
A Few Challenges
One challenge was the Moderator was not an expert on Calisti or the subject of modern applications. Even with a script, the flow of the discussion was still unknown and influenced by the expert's understanding of the questions. In some interviews, questions had to be rephrased and repeated to the expert. During the interviews, I would bring up questions the Moderator did not ask during the interviews and ask the expert follow-up questions to clarify some of their comments during the interview.
My unedited notes with time stamps
Sample of the recording transcript
Turning Text Into Data
I took detailed notes even though the interviews were recorded. I highlighted certain parts of the interview as significant information. I also used my notes to compare to the transcript. I did find some discrepancies.
I was responsible for analyzing all the transcripts using Dovetails as one of the research tools. I entered five of the six transcripts into Dovetail. I read these five transcripts to determine any misinterpretations or errors, making corrections as required. Because of privacy policies, the voice recordings were only available for about ten days before being eliminated.
Next, I tagged each transcript, highlighting specific subjects sentence by sentence. The tagging is critical in several ways. One is to be used to do analysis filtering, and the second is for Dovetail AI to generate an AI analysis summary.
Organizing the data
I organized the data using Excel by POC, Interface, and Gen AI. Each had its page with all six experts identified and data sorted by type. The data was more straightforward to compare to each other, and also what the conscientious was by type of data.
POC (Proof of Concept)
I compiled the POC learnings in a timeline graph to make consuming the formal POC process information learned from four expert transcripts easier to understand the flow and duration.
Harvesting the Data
Learning how companies execute POC (proof of concept) was valuable information not planned initially as part of the research. The first interviewer just so happened to reveal their process and the challenges that led up to their current POC process. From then on, asking about POC became part of the discussion topics. Four out of six companies had a formal POC process.
100% of the experts stated GitOps was the preferred Interface; next was CLI. "GitOps is the future." 100% of the experts indicated that GUI was the last Interface. "GUI is the last option." It was also noted that each has a place and set of users.
83% of the experts were not using Gen AI. 5 out of 6 had a favorable view of Gen AI.
"It's definitely empowering engineers and others."
"We don't want to bring in (Gen AI) in for the sake of it."
The research provided important indications about GenAI and the role each type of Interface has in their company. Learning about the formal POC provides insights we were not initially looking for. It is a plus and a good reason for researching to find hidden nuggets of information.
