The layer between a frontier model and a health worker's phone.
Crane builds reusable technical and institutional layers. The same capabilities serve healthcare, agriculture, education, finance and public services — documented once, here, and linked from each sector.
Crane builds reusable technical and institutional layers.
The same capabilities serve healthcare, agriculture, education, finance and public services — documented once, here, and linked from each sector.
Institutional integration
We embed AI into workflows, procurement, governance, training and evidence generation so pilots can become durable capability.
Deployment across the full spectrum
We engineer systems across the whole chain — climate and satellite-derived data sources, cloud and sovereign compute, edge nodes, and affordable phones running offline — rather than assuming permanent connectivity at any layer.
Evaluation and safety
We test performance, cultural context, domain boundaries, harmful outputs and real-world failure modes before systems move into deployment.
Speech and multimodal interfaces
We build speech recognition, text-to-speech and multimodal interaction layers that connect people to digital systems in the languages they use.
Models and language adaptation
We adapt open model families for African languages, local domains and constrained operating environments.
EaseHealth
|
Deployed
|
Controlled Field Deployment
Offline clinical decision support for frontline health workers, on Android, in local languages, with no connection required. Built on MedGemma, adapted to Uganda's clinical context.
how it works
INPUT
Health worker enters or speaks the clinical information
STRUCTURE
EaseHealth structures the case and surfaces red flags, relevant checks and areas requiring attention
BOUNDARIES
System displays confidence and stays within defined clinical boundaries
JUDGEMENT
Health worker investigates, applies judgement and decides
RETAIN
Assessment is retained locally and synchronised later where permitted
deployment detail
capability
publication gate
Primary users
Midwives, clinical officers and other frontline health workers.
Primary buyers & Institutional users
Health systems, ministries, hospitals, public-health programmes.
Deployment
On-device Android inference, offline; later synchronisation where enabled.
Clinical role
Decision support. The health worker remains responsible for assessment and action.
Clinical boundaries
- No autonomous diagnosis
- No replacement of clinical judgement
- Dosage recommendations deliberately excluded in the documented deployment
"It doesn't prescribe. It guides. It makes you ask the right questions."
⎯⎯⎯
Apollo Wamono
∙
Clinical Officer
Crane Voice Studio
|
Closed Beta
|
In Development
A workspace for capturing, transcribing, generating and integrating speech across African-language workflows. More than a model demo: a usable path from audio input to reviewed output and real deployment.
how it works
CAPTURE
Record or upload
TRANSCRIBE
Transcribe
REVIEW
Review and edit
OUTPUT
Generate or export
DEPLOY
Integrate
CAPABILITY & PUBLICATION GATE
capability
publication gate
Speech to text
Publish only with measured samples
Text to speech
Confirm quality, consent and availability
Speech workflows
Confirm end-to-end UI
Models and languages
Never claim universal African coverage
API and SDK
Developer preview or waitlist until docs, auth and versioning are stable
Deployment options
Do not promise on-prem or air-gapped without an implemented path
Consent and provenance
Required before any voice-cloning claim
proof required before launch
- ASR WER/CER by language and condition
- TTS listening evaluation
- Latency on named devices
- Supported-language matrix with sample audio
