Cases
A network of more than 1,500 offices sorts cells in blood smears automatically and highlights important zones in cytology. The laboratory doctor does the expert review, and verification stays with them. Another network in 90 cities gave the explanation of results to an ensemble of agents validated by experts: the patient gets the explanation with the result, free of charge (2025). A university hospital in the US trained a language model on 4.1 billion words of medical records and predicts 80% of readmissions.
The rule
Supporting, not replacing: this is how the laboratory network from the first case describes the role of AI. Any text that goes to a patient passes through a validating doctor. This rule removes both the legal and the reputational risk.
The service layer
Test preparation, booking a sample collection, result status, documents for corporate clients: this is the layer we build for laboratories in 14 days, with no medical device registration.