Built locally.
Connect across countries.
Public Services
Public institutions need AI capability they can govern, evaluate and operate — not systems they rent and cannot inspect.
What we are solving
Systems
Voice AI Pathways in Africa
Public-service voice interfaces
UCCB and cultural evaluation
Language adaptation
Trust and Governance frameworks
Crane Enterprise SDK
evidence
Programme
UNDP AI Hub programme work
Deliverables
Governance and strategy deliverables · public-service prototype workflows
Engagements
Institutional engagements under the G7-endorsed AI Hub for Sustainable Development
IRIS Africa
Local-language AI, voice technologies and inclusive digital access
Twist (Quadrivio)
Expressed interest in refurbished Android devices for offline voice AI deployment
Event
AI Hub Bologna Ecosystem Event, June 2026 — both partnerships signed at this event
2
PARTNERSHIPS
6
SYSTEMS
ACTIVE
STATUS
Partners
- IRIS Africa (Partnership signed · local-language AI, voice technologies and inclusive digital access)
- Twist / Quadrivio Group Industry 4.0 Fund (Partnership signed · expressed interest in providing refurbished Android devices for offline voice AI deployment)
- UNDP AI Hub for Sustainable Development (Programme participation · governance and strategy deliverables, public-service prototype workflows)
Healthcare
Frontline health workers make consequential decisions without specialist support, current guidance or reliable connectivity. Crane builds clinical decision support that runs where the cloud does not reach.
What we are solving
Systems
EaseHealth
Crane-Med
MedASR
TANO health workflows
Health evaluation and safety systems
evidence
Location
Luweero District, Uganda
Scale
15+ public health facilities · 268+ consented frontline health workers
Research design
A 50-CHW cluster-randomised study sits within the wider deployment, conducted with the Infectious Diseases Institute (IDI), Makerere University
Approvals
Makerere University School of Health Sciences REC MAKSHSREC-2026-41 · Uganda National Council for Science and Technology HS7408ES
Funder
Google.org Health AI grant
Evidence type
Deployment, feasibility, adoption, usability and safety — not clinical effectiveness
136
ENDLINE SURVEYS
81
ON-DEVICE ASSESSMENTS
0
ADVERSE EVENTS
Partners
Infectious Diseases Institute (IDI),
Makerere University - clinical research partner
Makerere University School of Health Sciences REC and UNCST - ethics oversight
Luweero District health authorities - field coordination
Google.org - funder
Financial Services & SMEs
SACCOs, micro-lenders and small traders run on records, conversations and field visits — in languages the tools do not support. Crane builds local-language voice and workflow infrastructure for that reality.
What we are solving
Systems
Crane Nemo ASR
Crane Nemotron reasoning layer
Crane Voice Studio
SACCO assistant (prototype)
evidence
Status
Pilot preparation
Approved wording
Crane is exploring local-language work-note and SME-finance workflows with a UMRA-regulated Ugandan lender.
Condition 1
Pilot agreement in place
Condition 2
Users onboarded
Condition 3
Usage logs exist
Condition 4
Naming cleared in writing
4
SYSTEMS
PILOT
STATUS
4
CONDITIONS
Partners
Crane is exploring local-language work-note and SME-finance workflows with a UMRA-regulated Ugandan lender. Conditions required: (1) Pilot agreement in place, (2) Users onboarded, (3) Usage logs exist, (4) Naming cleared in writing.
Education
Teaching tools assume English and assume connectivity. Crane builds foundational literacy and numeracy support for teachers and children in their own languages, on affordable devices.
What we are solving
Systems
Ganda-Gemma 1B
Bridge (FLN model)
Ganda-Gemma LiteRT
FLN training dataset
Luganda pre-training corpus
evidence
Completed outputs
Twelve completed technical outputs
Benchmarks
Two co-developed benchmarks (PCK + ELL)
Research
"Talking Teachers' Language" — research paper on AI performance in African languages for education
Teacher research
10 semi-structured interviews, audio recordings, transcripts teacher co-design research
Pilot
TANO Education Mode — 10 active pilot teachers in Uganda
Licence
All complete, open-source (Apache-2.0)
12
TECHNICAL OUTPUTS
10
PILOT TEACHERS
2
BENCHMARKS
Partners
Fab AI Foundation / AI-for-Education.org (Named technical partner for their AI-for-Education programme; Fab AI Foundation backed by the Gates Foundation and UK FCDO — Grant P2/W1/06 completed) · Google DeepMind (FLN report: 'Sovereign AI for Foundational Literacy and Numeracy') · All complete, open-source (Apache-2.0)
Agriculture
Advisory information rarely reaches the farmer in the right language, at the right moment, on the device she owns. Crane builds voice-first, weather-informed advisory that works on 2G and offline.
What we are solving
Systems
Swahili-Gemma
Farmer records
WeatherNext integration
Africa's Talking (SMS/USSD)
K2-Shamba crop diagnosis
On-device inference
evidence
Shamba Records
Agricultural data platform (~50K farmers), carbon, insurance and farm-data layer
Hello Tractor
Mechanisation and booking-agent network
WeatherNext (Google DeepMind)
15-day forecasts, 0.25 resolution
Africa's Talking
SMS/USSD messaging infrastructure for 2G reach
Kayebwe demonstration
Offline Luganda interaction on a low-cost phone in airplane mode
World Bank
"Harnessing AI for Agricultural Transformation" case-study feature
7
PARTNERS
~50K
FARMERS REACHED
5
SYSTEMS
Partners
Shamba Records (MOU signed · Agricultural data platform (~50K farmers), carbon, insurance and farm-data layer. Introduced via Google Research Africa) · Hello Tractor (MOU signed · Mechanisation and booking-agent network) · Google DeepMind (WeatherNext: 15-day forecasts at 0.25° resolution, access confirmed) · Africa's Talking (Component: SMS/USSD messaging infrastructure for 2G reach) · Kayebwe demonstration (Field demonstration: Offline Luganda interaction on a low-cost phone in airplane mode) · World Bank (In discussion: 'Harnessing AI for Agricultural Transformation' case-study feature)
