Launching a Telemedicine Platform During a Healthcare Crisis

In April 2020, the question was not "should we build this?" It was "can we ship something useful before the worst week?"

The client was a healthcare group whose physical clinics had emptied within days of the lockdown. Patients still needed care; doctors still wanted to provide it. The gap was the access layer. We had to build it under time pressure that did not forgive over-engineering.

Context

Telemedicine is a small problem on the surface - patient books, doctor consults, prescription emerges. The complexity sits in the trust layer underneath. Patients have to believe they're talking to the actual doctor and that what they hear is going on a record. Doctors have to believe their workflow inside the platform won't slow them down compared to in-person practice. Regulators have to believe the platform handles patient data responsibly.

In April 2020, none of this had public reference designs in our market. The platform had to be opinionated and fast, both at once.

The product challenge

Three flows had to land cleanly on day one - patient appointment booking, doctor consultation, and post-consultation prescription / record. Anything that broke trust in any of those three flows poisoned the whole thing. Patients who had a glitchy first call did not come back; doctors who hit a snag refused to do the next slot; either side bailing out cascaded.

So the engineering bar was not "feature-rich." It was "the three flows work, every time, with low-bandwidth tolerance, and create a record that satisfies a clinician reviewing it later."

My role

I led product and engineering for the consumer side and the doctor workflow. Architecture, the patient mobile and web experience, the doctor's consultation interface, video integration, prescription generation, secure record storage, and the ongoing patch cadence as we learned where the trust failures actually were.

Core features

• Appointment booking: doctor availability, slot selection, payment, confirmation, calendar reminders.
• Pre-consultation flow: symptom entry, history, file upload, so the doctor walked into context, not a cold call.
• Video consultation: integrated, low-bandwidth-tolerant, with mid-call escalation to audio-only when video failed.
• Doctor workspace: patient queue, current case, history of past visits, prescription pad, follow-up scheduler.
• Prescription delivery: generated as a record, signed, delivered to the patient with attached care instructions.
• Audit trail: every appointment, consult, prescription captured for clinical and regulatory review.

Technical highlights

Video was the single highest-risk technical surface. Most patients were on mobile, on intermittent home Wi-Fi or 4G. Most doctors had moved to home offices with whatever ISP they had personally. We integrated a managed real-time-comm provider rather than rolling our own - the wrong place to optimize cost when "must work" is the requirement.

Below the video layer, every consultation produced a structured record - patient ID, doctor ID, time, files exchanged, prescription issued, follow-up scheduled. The record was the source of truth, not the call. Even if the video had glitches, the consultation as data still existed cleanly, which mattered to the clinician's record-keeping obligation.

Prescription delivery was where most "telemedicine MVPs" cut corners. We treated it as a real document - signed, formatted, delivered through a channel the patient already trusted (WhatsApp / SMS / email per preference), with a copy-back into the patient's record in the platform. Doctors stopped issuing handwritten prescriptions over WhatsApp. That single change made the platform clinical-grade rather than improvised.

The doctor's workspace was deliberately spare. Patient queue on the left, current consultation on the right, history one click away, prescription pad always reachable. We rejected every feature request that added a step. Doctors gave us back time when we didn't waste it.

What this taught me

Speed without breaking trust. This phrase became the operating constraint. Speed meant patients could get to a doctor when their physical clinic was closed. Trust meant the consultation produced a record a clinician would stand behind. Optimizing one without the other would have killed the platform inside a quarter.

Healthcare apps can borrow from consumer products - but only the surface, not the substance. The booking flow looked like a consumer app. Underneath, the data model and the recordkeeping had to satisfy clinicians and regulators. The product was a consumer wrapper over a clinical-grade core.

Reduce anxiety in every flow. Every screen had a state - confirmed, in-progress, completed, prescription-ready. The state was always visible. Patients waiting for a video call to start did not have to wonder if it was happening. Anxiety reduction was not soft UX; it was the single most-cited reason patients used the platform a second time.

Outcome

The platform absorbed clinic-scale consultation volume during a period when in-person care was impossible. Doctors stayed productive without commuting; patients reached care without leaving home. The structured record layer made the platform credible to clinical reviewers and regulators after the immediate emergency passed. The trust hurdle - usually the slowest part of healthcare adoption - fell faster because the alternative was no care at all.

---

If you are building under time pressure where lives or livelihoods are touched: ship the three flows that matter, do not invent features, treat the audit trail as the platform. The polish comes after.