Photo by from Pexels

What does “tool compound”, “target expression”, or “IC50” mean? It’s fine if you don’t know, I didn’t either when I started working at Benevolent as a Product Designer. For this reason, I work with our expert users to discover drugs.

They help to understand the domain and, more importantly, shape the requirements of digital products. All Benevolent products are co-created by technologists and scientists working closely together. Why? Without knowing and understanding your users’ needs, it is impossible to design a good product. BenevolentAI is a product design company that focuses on designing, usability, and utility of our end-to-end AI drug discovery platform. I aim to develop digital products that enhance scientific research at various stages of the drug discovery process.

Our products must fulfill some very specific scientific requirements in order to meet this objective. Users of our Platform, for example, handle a great deal of information. Information about a particular segment of biology may help paint a coherent picture, or it may present conflicting or contradictory evidence. Due to the complexity of biology and the fact that it is still poorly understood, biomedical data is equally complex and incomplete. New software products enabling AI-enabled drug discovery should enable our scientists to navigate this complex data space in order to make unbiased, better informed, and data-driven decisions.

BenevolentAI’s scientists worked with our internal users to meet this challenging design brief.‍

I collaborate closely with Product Managers and Software Engineers at BenevolentAI, as well as with a variety of scientific experts, including Drug Discoverers, Medicinal Chemists, Bioinformaticians, and Cheminformaticians. Fortunately, these users are also our colleagues, and we work closely with them every day. I learn about their thought processes, their motivations, and their expertise from them; they are not just users, but also collaborators. My close relationship with them enables me to deliver designs that help them discover new and better drugs for patients in need without having any scientific background.

It can be challenging to design for and collaborate with expert users. The following is a description of my design process to help other designers who are working in a complex, collaborative scientific environment‍

The first rule of design is not to panic, and to take each step at a time. When you don’t know everything right away, it’s OK, and even encouraged, to ask questions. Benevolent scientists are passionate about their work and field of expertise. They are able to convey difficult topics in a way that is easy to understand for outsiders. As a team, we are used to collaborating on solutions, since nobody has all the answers. In this regard, we can embark on a 3-part process; research, design, and consultation, followed by usability testing.

  1. Research

Our design team conducts research based on the scope and complexity of each project; it is conducted by user researchers, designers, and product managers. I conduct research on some projects and utilize the findings of our researcher on others. Observation is one of my favorite research and empathy-building techniques.

As our users are in-house, we are able to schedule one-hour sessions with them to observe how they use our products, how they work, what their challenges or pain points are, etc. Your design decisions will be better based on such observations. I observe our user researcher conducting user interviews and take notes. User interviews are another useful research method. As a result of the interviews, I receive a “Proposed requirements” document that helps me begin the project. It is sometimes helpful for me to watch the recordings of the interviews again (if the domain is new to me), to understand who our users are.

I conduct user interviews myself when a project does not require the attention of the user researcher. The smaller the project, the fewer meetings I organize in order to get input from the experts and to gain a better understanding of the topic. In order to get a variety of viewpoints, I try to keep the pool of experts diverse. I learned about the drug discovery process during the onboarding process, so I was familiar with the domain of my users. To get some previous knowledge before speaking to users, it is also a good idea to read about the domain yourself.

I start developing initial designs once I have gathered basic information about the topic and discovered the users’ pain points.

  1. Consulting and design

My design process involves both design and expert scientist evaluation. Before I evaluate designs from a usability perspective, I get feedback from my domain expert colleagues to get a scientist’s perspective.

My first step is to design a low-quality or high-quality model, depending on the situation. I always check my designs with experts and ask for feedback: is everything right, is something missing? Before the usability check, this is important, otherwise scientists are more likely to focus on details than on the design, such as irregularities in dummy data instead of performing activities. I also try to spot any usability problems the designs may have (although it is not a priority at this stage).

We try to keep the group of experts as diverse as possible and separate those taking part in the design process from those conducting usability checks. This allows those in the latter group to see the designs for the first time when conducting usability checks. During the final phase of the design process, other stakeholders, such as software engineers and product managers, are involved to learn about different perspectives. It is an iterative process to design, show the prototype to users, and everything happens simultaneously. Despite working closely with users, I still need the fresh perspective of my primary users who are not involved in the project. To eliminate the implication that designs are being tested, I organize user testing sessions, or “usability checks”.


Using prototypes to test usability‍


Depending on the project, the usability check takes different forms. When working on smaller projects, I ask users to perform a few activities on the interactive prototype during a briefing. Based on how the designs perform, I make adjustments. If I have time, I might also consult expert users about the design. You will be able to gain new insights from domain experts who were not involved in the project before. My ability to find small details or requirements that are missing increases the more people I ask. Due to the fact that I observed the users while they worked during the research phase, I was able to prioritize my findings.

    We hold longer usability checks for bigger projects.  In case I need more input from a diverse group of users, I set up a remote user testing session (with that allows users, on their own time and terms, to provide feedback on the usability of the site.  As I don’t have face-to-face contact with the experts, I use Maze at the end of the process, or when the nature of the design is more focused on interaction design and doesn’t require much expert input.

    Whenever possible, I use interactive prototypes and test them in users’ normal workflow in cases of complex interactions or hypotheses. A key flow in Drug Discoverers’ workflow had to be redesigned for one of the first large projects I had at BenevolentAI. My idea was to improve how Drug Discoverers collected their assessments about genes, and I was interested in finding out how users would react. Despite the fact that the design wasn’t as complex, the new flow wasn’t the same as what users are used to. To simulate new product features, we used Google Sheets with pre-formatted functional cells. In the prototype setup, BenevolentAI’s platform took up 70% of the screen, while the Google Sheet prototype took up the other 30%. I thus had the opportunity to observe how users interact with the proposed design, and the cost of prototyping was very low.

Comparing this prototype to other quick and easily accessible prototyping tools, I appreciated the interactivity and the ability to accept user input. By using this approach, I could: quickly and efficiently confirm my hypothesis, see how users react to the new flow, and find potential problems with the design with just a few cycles of usability tests. Whenever necessary (such as when I find some usability issues with the design or something misleading in the design), I redesign and repeat the usability checks.

The 3 main steps of my design for and with experts are described above. The process is highly interactive, and the details vary based on the project, the users, the timeline, and the implementation.

It can be challenging for designers without scientific backgrounds to work at a company with such a highly scientific business domain. In the domain of drug discovery, constant collaboration with our users is crucial to the development of visualizations, dashboards, and design tools.


Leave a Reply

Your email address will not be published. Required fields are marked *