Most AI tools assume data can be sent wherever the model runs. But Indonesia's Ministry of Health expects health data to stay within the country. So we built the system differently. The assistant runs on premises or in a private cloud hosted in Indonesia, which means patient data never leaves the country.

People often assume local AI is too slow compared with overseas APIs. We solve that by running NVIDIA GB10 devices inside the hospital, keeping inference fast without compromising compliance.

The assistant handles repetitive tasks like drafting clinical notes, summarising patient histories, and bringing up relevant context before rounds and handovers. It does not replace clinical judgement. It simply gives doctors and nurses more time to focus on patients instead of paperwork.

Agile D3velopers needed a working private blockchain MVP that could serve as a trust layer for the Janjilah App without over-engineering the first release.

I designed the private blockchain architecture, developed a production-grade network, and deployed it so it could be integrated into the app.

The MVP delivered a working blockchain foundation that supported Janjilah App's core trust and verification workflows, ready for further product iteration.