Centyent AI
Intro
Centyent is working on an AI-powered clinical reasoning product that is designed to integrate with existing healthcare systems, supporting clinicians in making faster, more accurate diagnoses and ultimately improving patient outcomes. We were brought in to research the space and present our findings and protoype suggestion back to the business. Our starting point was understanding the reality of clinical work - how clinicans move through diagnosis, treatment, and documentation, where the friction lives, and where an AI assistant could add genuine value rather than add noise to an already demanding workflow. To answer that, we ran a research sprint that combined clinician interviews, stakeholder sessions, competitor analysis, and desk research. The findings shaped a prototype that we presented to the business, and the in-house product team used for user testing and compared with their findings.
User Experience
Product Design
UI/UX Design
User Interviews
User Flows
Prototyping
Year
/
2026
Client
/
Centyent AI
Recruiting clinicians for research presented a significant challenge, their time is exceptionally scarce and their schedules largely inflexible. Despite this, we secured three in-depth interviews and one conversational interview. What emerged was a nuanced picture of how diagnostic reasoning actually works in practice. It's rarely a linear process - clinicians draw on their own accumulated knowledge, consult peers and seniors, and layer in independent research, often simultaneously and under significant time pressure. Diagnosis is as much a social and experiential practice as it is a clinical one.
It also highlighted the risks around introducing a new AI tool into this environment. Trust in AI, adoption barriers, and onboarding friction all came up repeatedly. Clinicians are already navigating system overload and information noise, and anything that adds to that burden risks being ignored or abandoned entirely.
These insights crystallised our design challenge and problem statement: How can we design an AI clinical support tool that improves diagnostic accuracy and supports professional learning, while preserving the human element of decision-making? And how might it integrate frictionlessly into the systems and rhythms of everyday clinical practice?
We synthesised our findings through user personas and empathy mapping, drawing out the key themes and patterns emerging from our research. These informed a round of ideation and concept development, grounded further by desk research across medical papers and clinical studies to pressure-test our thinking. From there, we moved into prioritisation, defining the features most likely to address the core clinical needs we had uncovered, before translating these into two prototypes. The first explored a second-opinion AI model combined with a case-sharing community, giving clinicians a trusted space to sense-check decisions and learn from peers.
A key insight from our research was that adoption and onboarding posed one of the greatest barriers to introducing a new tool into clinical workflows. Rather than asking clinicians to learn an entirely new interface, we explored a different approach - what if Centyent could adapt to the clinician, not the other way around? During onboarding, users would select the EMR system they already work with, and Centyent would mirror that familiar UI, layering its own features: case sharing, second opinion AI, and community - on top of an environment clinicians already know and trust. For our prototype, we replicated the Cerner interface, as it was the system most commonly used by the clinicians we interviewed.
The second was an early concept for a mobile-first learning tool, designed to work in tandem with the diagnostic findings and support ongoing professional development. Both prototypes were presented to the Centyent team alongside our research and rationale. The next step is for their in-house product team to develop the concepts further and conduct user testing with clinicians — validating our findings and continuing to shape the product with real-world insight.
