Nambi - our teacher - is an AI that runs entirely on a $50 device, with zero internet. Today she teaches ICT for $1.30 a student. Tomorrow the same engine carries health guidance, farming advice, civic info, and the first AI that thinks in Luganda, Swahili and Acholi - off-grid, off-cloud, on the continent's own terms.
The curriculum exists. The timetable exists. The hardware, sometimes, exists. What doesn't exist is the qualified teacher, the Wi-Fi, or the $80/month budget for a platform that assumes all three.
Qualified ICT teachers are rare across Ugandan secondary schools. Most timetables list a computer lesson with no one qualified to deliver it.
Average individual income. Mobile data takes a significant share. Every existing EdTech tool assumes a monthly subscription most students cannot pay.
Of Ugandan youth are unemployed or underemployed. Employers consistently cite lack of basic digital skills as a primary barrier to hiring.
Much of Uganda has no reliable internet and intermittent power. Every cloud AI stops the moment the connection drops. Most of the school day.
Omuntu w’omuntu. A person is a person through other people - and a technology is ours only when it works in our language, on our hardware, at our cost. ICT today; health, farming, civic services tomorrow.
Nambi is not a chatbot wrapper. It is a full on-device AI tutor, a classroom data engine, and the seed for an African-language foundation model - all in one programme, built to scale horizontally without infrastructure as a prerequisite.
Nambi runs a genuine large language model entirely on-device. No cloud, no API calls, no connection. The AI loads, reasons, and responds on hardware that costs less than a month's salary for most Ugandan workers. This is not a quiz engine.
Every question asked, every answer given, every correct and incorrect response is logged locally. Aggregated and anonymised, this creates something that has never existed before: a dataset of how African students learn, built from African classrooms.
Current AI models are trained almost entirely on English. Nambi's classroom interactions are generating training data for the first AI that genuinely thinks in Ugandan languages - not translated from English, but built from Uganda.
Nambi's offline AI architecture is subject-agnostic. ICT skills today. Vocational trades tomorrow. Same engine, different curriculum. Same model, different language. Built to replicate across East Africa without infrastructure as a prerequisite.
Word, Excel, data entry, email, digital filing. The exact skills Ugandan employers list in job adverts. Testable, practical, taught through conversation - not textbooks that age on a shelf.
Instant explanations, local logs for teachers, per-student progress for administrators. Identifies exactly where each student is struggling - without a qualified teacher physically present in the room.
The model runs entirely on the device. Once set up, Nambi works at any power level, with zero data cost. One device. One programme. Indefinitely - no subscription, no surprise bill.
Every session is logged automatically. Dashboards for programme managers. CSV export for grant reporting. All generated locally - no extra software, no internet connection required.
Every claim on this page is traceable to field data, registered entities, or a named partner. Full pack (methodology, session logs, outcome metrics, audited accounts) available on request to funders.
Nambi proves itself teaching ICT. But the thing we are really building - low-powered, offline AI, on hardware Africa already owns, in languages Africa already speaks - does not stop at the school gate. Every sector that was told to wait for better internet can stop waiting.
Community health workers, kilometres from any signal. Offline AI on a $50 phone triages symptoms, doses paediatric medicine, flags danger signs in pregnancy - in Luganda or Swahili, without a data bundle.
Smallholders losing harvests to pests they cannot name, at markets they cannot price. Offline AI answers what is eating my maize and when do I plant - in the farmer's own language, on a feature phone.
Land titles, benefits, tax forms, legal rights - explained in plain Acholi or Swahili, at the sub-county office, without the query routing through Nairobi or London. Civic AI that works where the state actually is.
Millions of African adults never learned to read in their mother tongue, let alone English. Offline AI is the first scalable 1-to-1 tutor that can sit with them - patiently, for the hour it takes - without a class, a teacher, or a data plan.
Electricians, mechanics, tailors, masons. Offline AI explains a wiring diagram, diagnoses a gearbox, teaches a new pattern - in the back of a roadside workshop, with no Wi-Fi and no excuse.
Each sector above, trained on African data, in African languages, running on African-owned devices. Not an API rented from a foreign hyperscaler - infrastructure the continent can actually own, operate, and export.
ICT skills training with Nurture Africa VTC. 150 students, live. $1.30 per student per term. Proving the hardware, the offline AI, and the unit cost - the wedge that unlocks every sector after it.
Same engine, new curriculum. Offline triage for community health workers. Pest, weather and market guidance for smallholder farmers. Vocational trades. The subject changes; the cost model holds.
The first foundation model trained on real African classroom and field data. Luganda first, then Swahili, then Acholi. AI that stops translating from English and starts thinking in the languages of its users.
Nambi becomes the offline AI layer for the Global South - education, health, agriculture, public services - any subject, any language, any $50 device. Deployed without waiting for better infrastructure.
Akaalo is registered in Uganda (2023) and Ireland (2025). Actively seeking grant funding, research partnerships, and hardware support to scale.
Per-student cost ($1.30/term), session logs, quiz deltas, and 8-week outcome metrics from our live pilot. Full evidence pack, unit economics, and entity docs on request.
Every session generates grounded interaction data from real African classrooms, in African languages, on offline devices - a corpus no hyperscaler can scrape and no foreign lab can replicate. The substrate for the world models the Global South will need: learning today; health, farming, civic life next.
Nambi runs on low-powered laptops and tablets with 4GB+ RAM, from $50 upwards. 20 devices can run a full cohort. No ongoing internet costs - hardware is the scaling constraint.
Not a subscription. Permanence. Curriculum, device time, and a local operator paid fairly - for one learner, for one full school term. No renewal. No recurring cost. For businesses, sponsor a cohort or a whole school and see your brand on the devices your money put in the room.
Named cohorts, co-branded devices, quarterly impact reports, and a site visit to Nansana. Starts at $1,950 / cohort. Talk to us - info@akaalo.com.
Secure payments via Stripe and PayPal. Akaalo Ltd, registered Uganda and Ireland. Grant & partnership enquiries - info@akaalo.com
Whether you are a funder, researcher, school, or hardware partner - we want to hear from you. Real humans reply within 2 working days.