Inky HealthMedical AI, built by people who understand medicine.
Inky Health trains models on clinical and research data, builds private AI and document systems, and advises on what is worth building. We work with clinicians, researchers, hospitals, and the government teams that support them.
What we do.
Four ways in. Most work starts with one and draws on the others.
Model training
Language and vision models fine-tuned on clinical and research data, evaluated against clinician-labelled ground truth.
Private AI
Search and question-answering over guidelines, policies, and records, with citations and models that run inside your network.
Document systems
OCR, extraction, and redaction for scans, forms, tables, and large document collections.
Advisory
Feasibility, data readiness, regulatory path, and build-versus-buy decisions before a budget is committed.
Models trained for clinical work.
The kinds of models we build with clinicians and researchers.
De-identification
Models that find and remove personal information in clinical notes, letters, and records, tuned to local formats.
Diagnostic signals
Models trained on laboratory, pathology, and imaging data to surface findings earlier for clinical review.
Decision support
Research models that learn from clinician decisions, with the evidence trail ethics committees and regulators expect.
Clinical text
Summarisation, coding, and structured extraction from notes, reports, and referrals.
Built to stand up to review.
Measured before release
Evaluation sets are agreed with clinicians before training starts, and results are reported across patient subgroups.
Data stays inside the boundary
Training and inference run on your infrastructure or ours, designed for Australian data residency.
Ready for review
Documentation that supports HREC applications, privacy assessments, and TGA conversations.
Start a conversation.
Describe the model, system, or decision you need to move forward, and the data you have to work with.
Enquiries are typically answered within one business day.