Automating the medical concierge

DoctorAnywhere (DA), Southeast Asia's top telemedicine provider, has clinics across Singapore. When a GP refers a patient to a specialist, the medical-concierge team (MCT) helps book and confirm the appointment.

Contact Center Management (CCM) centralizes all patient interactions, making a contact center easier to manage and operate. By streamlining the GP-to-specialist consultation process, CCM reduces costs and lifts customer satisfaction.

The MCT tracked every patient request inside a complex Excel sheet — fragile, manual and impossible to scale.

• High human-error rate from updating all requests in one complex spreadsheet.
• An average of 32% daily user complaints.
• 48% patient drop-off due to the time MCT took to complete a case.
• 89.5% of appointments couldn't be secured and proceeded.

Success metrics

• Appointment-request conversion: how many requests convert to confirmed appointments.
• Complaint rate: reduce daily patient complaints.
• Digitalise the manual flow: cut human error and time-to-complete per case.
• Support requests: reduce app-related support volume.

For most product decisions a hybrid approach works best: build a current-state map to understand existing opportunities, then a future-state map to envision new ideas holistically. Use a hypothesis-first map to build stakeholder buy-in, then follow up with user research to validate or evolve those assumptions into a second, evidence-based version.

• Mapped the current GP → specialist referral flow.
• Mapped the current GP → specialist cashless-insurance flow.
• Analysed the MCT team's data-collection Excel document.
• Understood current patient sources and escalation paths.

From the current workflows I ran a workshop with stakeholders and created a hypothesis map across the key stages — case creation, case assessment and hospitalization — and validated it with them.

I then prepared a 15-question interview script focused on values, pain points, motivations and daily routines, and ran face-to-face interviews with 3 medical-concierge team members and 2 stakeholders — refining the hypotheses and assumptions against what I heard.

Current-state journey maps visualise the experience customers have trying to reach a goal with the product as it exists today. Future-state maps visualise the ideal journey that doesn't exist yet — the target we designed toward.

Before any pixels, I developed the end-to-end user flow and validated it with stakeholders and the medical-concierge team — confirming each decision point, status change and handoff held up against how cases actually move.

With the flow locked, in Miro I translated first sketches into low-fidelity wireframes, ran a moderate usability test with stakeholders, and used the findings to enhance the lo-fi and validate again before locking the final version.

Using the design system, I built high-fidelity prototypes that streamline how agents manage and track cases. Each case contains multiple tickets — appointment requests or general inquiries — each with its own tailored workflow.

For appointment requests, agents follow progress through defined status stages and update details from patients and specialist clinics, covering the full GP-to-specialist flow designed for the MVP.

I ran in-person unmoderated usability testing across 8 tasks, observing behaviour, collecting feedback and completing iterations. At the end, participants filled out the System Usability Scale.

Next, we started the following phase — adding the de-prioritised “could-have” features and working closely with the data team to find behavioural gaps and improve them.