There is a recognisable lifecycle to a great many technology-for-development projects. A pilot is announced with a partner and a funder. It runs for twelve to eighteen months. A report is published describing promising results. Then nothing further happens, and two years later a similar pilot is announced by someone else.
The funding cycle rewards starting
Grant funding is generally allocated to projects with a defined beginning and end, and reported on at the end. Novelty is rewarded in applications; continuation of somebody else's idea is not. That produces a strong incentive to launch something new and almost no incentive to maintain something that already works.
Maintenance is where the actual value sits. A system that runs for a decade is worth far more than five systems that each ran for eighteen months, and it is much harder to get funded.
Handover is usually an afterthought
Pilots are frequently built by external contractors, on infrastructure the receiving institution does not control, with knowledge that leaves when the contract ends. Even a genuinely successful pilot then cannot be sustained by the ministry or hospital it was built for, because nobody there can run it.
The projects that survive are consistently the ones where local staff built or co-built the thing, where the code and the credentials transferred, and where somebody's ongoing budget line covers hosting.
A fairer test
Before funding a pilot, ask who will operate it in year three, whose budget pays for it, and what happens to the data if it stops. If those questions have no answers, the pilot is a study, and it should be described as one rather than as a solution.