Why Nepal.
Nepal is not the demo. Nepal is the proof — that a nation of thirty million, with no GPU cluster and no foreign vendor, can deploy sovereign Superintelligence at the electricity budget of a small office. Serve Nepal end-to-end, and the playbook exports to five billion people the frontier labs are not chasing.
The four constraints that make Nepal the ideal testbed
1. Bilingual constraint. English and Nepali, side-by-side, in every citizen interaction. If your AI can only serve one language, it cannot serve Nepal. That constraint alone breaks 95% of frontier-lab deployments before they start. AlifZetta was built bilingual on day one because it had to be.
2. Three-tier government. Federal, provincial, and local — 77 districts, 753 local governments, seven provinces. Every citizen query has to route to the right tier of authority, in the right jurisdiction, with the right form. That routing complexity is exactly what LATTICE was designed to solve.
3. Seasonal disaster cycle. Monsoon floods (June-September). Earthquake risk (year-round). Dengue outbreaks (post-monsoon). Landslides. Cold-wave mortality (winter). The AI has to know today's weather advisory, today's hospital capacity, today's snake-bite antivenom stock. That is what a 30-minute refresh substrate is for.
4. Measurable healthcare gap. Nepal's maternal-mortality figures, its NCD burden, its rural-facility staffing shortfall — every metric is public and every metric has room to move. If AlifZetta shifts even a percentage point on maternal mortality by giving skilled birth attendants better tools, that is a nation-scale outcome.
Why the playbook exports
Once you can serve Nepal — bilingual, three-tier, seasonal-critical, healthcare-heavy — you can serve any emerging economy. Bangladesh. Sri Lanka. Cambodia. Kenya. Nigeria. Peru. Bolivia. Rwanda. The same architecture, the same substrate model, the same CPU-first economics. That is five billion people the frontier labs are not seriously chasing, because their unit economics require thousand-dollar-per-month floors.
AlifZetta's floor is forty-seven dollars per month per production node. That flips the addressable market from twenty percent of the world to a hundred percent. Nepal is the working proof that the flip is real.
What we have already shipped in Nepal
- 94,000+ typed NEXUS entries covering 77 districts, 7 provinces, 20 cabinet ministers, 400+ government services, 300+ clinical pathways, hospital directory, university directory, media directory
- Live at demo.axz.si — sub-5ms retrieval, bilingual answers, cited responses
- Daily-pulse scraper — 30-60 fresh NEXUS entries per day: news, weather, quakes, FX, Wikipedia trending
- Nepali BS calendar handler — live AD↔BS conversion, tithi/nakshatra
- Emergency handlers — 16 structured first-aid cards including Nepal-endemic snake bite protocol (Krait, Cobra, Viper stratification), rabies category triage, altitude sickness
- Government API partnership — deployed at the OPMCM tier and used inside the ministry system
Why Padam is here
Padam Sundar Kafle is Nepali. Kathmandu-born, Kathmandu-headquartered, Nepal is the country he owes his engineering career to. AlifZetta is his answer to the question every diaspora engineer asks eventually: what if I built the thing my own country needs, from my own country, with my own people? Twenty-one years of engineering across thirty-five countries, and the most important deployment is the one twenty minutes from his childhood home.
See it running
Live at demo.axz.si — try a bilingual query, read the cited response, verify the sub-5ms latency yourself.
Try the Nepal deployment → Read the manifesto →