The minimum viable prescription platform
The difficult part of scoping an Rx launch is deciding what can remain manual and what cannot be optional.
A prescription platform can begin with a small catalogue, a few clinics, and staff reading each prescription themselves. It can defer automated extraction, deep integrations, and self-service onboarding. The customer can still receive the medicine through the required review and dispensing process.
An approval gate is different. Remove it and the transaction may proceed without the decision that permits it. Remove the connection between an action and the person authorized to perform it, and the platform can no longer establish who made that decision. A smaller launch does not resolve either problem.
For every proposed deferral, ask what will happen to an actual order without it. A manual substitute may make the journey slower or more expensive while preserving its integrity. A missing control may make the journey unacceptable. Those consequences matter more than whether both items happen to be called features.
In Platform Rx, we deferred several forms of automation and kept the full path from evidence to delivery. Engineering tested whether an order could advance without the required authority or record; participating clinics tested whether their staff could actually complete the work. A system that blocks every questionable order but offers no workable route to resolve it will leave customers waiting.1
The pilot should then reveal which manual work is worth automating. In our case, catalogue corrections supplied data for improving classification, while review time and workload helped determine when expansion was appropriate. Automated refills were deferred because generating more requests ahead of review capacity could worsen access.2
I would use the first release to discover where automation can remove work from a transaction already functioning correctly. That gives the next investment a concrete target and preserves a way to tell whether it helped.
Footnotes
-
Author’s Platform Rx product and system-design record, August 2026, page 2, V1 scope, engineering tests, and design-partner validation; pages 3 and 8, launch gates and deferred capabilities. The prioritization argument is the author’s interpretation. Back to reference 1
-
Design record, page 9, assisted classification and review-capacity trade-off; page 8, automated-refill deferral. Improvements to classification are described as a subsequent roadmap direction, not a measured result of the pilot. Back to reference 2