Awesome Digital Public Infrastructure

Infrastructure before applications — lightweight tools outperform complex HMIS in fragile states.

Ministry buy-in is the bottleneck — stakeholder alignment takes months, not weeks.

Consistent identifiers from day one — cycle ID mismatches break cross-system reporting.

People first, always — Human-Centered Design (HCD) is not optional. Run user research before writing a single requirement. The focal point filling in a form at 3G speed in a municipal office is your primary user, not the minister who will see the dashboard.

Service blueprints before system design — map the as-is service end-to-end before touching any technology. A one-day workshop with frontline staff reveals more than a month of requirements documents.

User journeys expose the real pain — walk the user journey for each persona. The to-be journey is your north star. Every technical decision should map back to it.

TOGAF helps frame the architecture conversation — even a lightweight ADM (business → data → application → technology) stops teams from jumping straight to tools.

Design offline-first, not offline-compatible — offline-first means every interaction was designed assuming no connectivity. The last-mile problem is political, not technical. Degrade gracefully: offline collection, async sync, SMS fallback, paper backup.

Drones, AgriTech, and AI follow the same rule — the question is never "is this technology exciting?" It is "does this solve a real problem and can it be maintained by the people who inherit it?" Most AI failures in development contexts are data failures, not model failures.

The digital divide is replicating, not closing — the AI divide compounds this: countries without clean, governed data cannot use AI-enhanced public services. Gender is a design variable, not a disclaimer. Build for the hardest-to-reach user first.