Twice as Fast: AI Helps Insurers Review Physicians’ Orders
An AI assistant developed by Ingosstrakh has cut the time needed to search clinical guidelines for voluntary health insurance cases in half.
Patient insurance has long been an integral part of the healthcare system. It helps ensure that people do not have to face illness alone: patients receive care, and healthcare providers get paid. But every insurance agreement involves a complex process behind the scenes.
When reviewing requests for services covered by voluntary health insurance (VHI), a medical reviewer must consider several factors. Does the prescribed service comply with clinical guidelines for the patient’s specific condition? Is the service actually necessary? Does it duplicate a test or treatment the patient has already received? Answering these questions requires reviewers to work with large volumes of information. The Russian Ministry of Health’s clinical guidelines database contains more than 700 documents, which are updated regularly.
To simplify this process, Ingosstrakh developed an AI assistant for its medical review team. The solution runs on the company’s proprietary IMAN platform and uses retrieval-augmented generation (RAG), a technology that generates answers based on information retrieved from relevant sources.
How the Assistant Works
The team built a knowledge base for the AI assistant using clinical guidelines. The system receives information about a patient’s symptoms, diagnosis and age, along with a list of prescribed medical services. It then identifies relevant sections of the guidelines and generates an assessment explaining whether the requested service can be justified under those recommendations.
This means medical reviewers no longer have to search for the relevant information manually every time. The assistant gathers material specific to each case, allowing them to move more quickly to the expert assessment.
According to Ingosstrakh, the assistant cuts the time needed to search for medical information in half. A centralized knowledge base also helps standardize the review process: reviewers draw on the same set of clinical guidelines and can apply a consistent approach when evaluating physicians’ orders. The AI does not replace clinical experts, however. It takes on a substantial share of the routine work.
A Focus on Productivity
The new AI algorithm is part of Ingosstrakh’s AI transformation program for its medical insurance division. The company sees artificial intelligence as a way to support professionals in processes that require them to handle large volumes of information while maintaining expert oversight.
“We are systematically reviewing our medical insurance processes to identify where AI can deliver measurable results. In this case, we are already seeing a substantial reduction in the time needed to find the information we need. The next step is to scale this approach to other processes while maintaining work quality and oversight by our specialists,” said Anastasia Skurikhina, director of Ingosstrakh’s medical insurance division.
New Opportunities for AI
Other Russian companies are developing similar solutions. In December 2025, Sovcombank Insurance announced the deployment of an AI agent to review requests for services covered by voluntary health insurance. The agent analyzes customer inquiries, medical records and policy terms, then recommends a decision. During the rollout, a human operator reviewed the system’s output. According to the company, processing time has since fallen fivefold.
In late 2025, AlfaStrakhovanie also launched a generative AI service for medical review. Powered by the Qwen language model, the system structures documents, compares requested services with clinical guidelines and standards, and flags potential discrepancies. The company reports an accuracy rate of 90% for the model in medical review.
Russian IT company Mains Laboratory is also contributing to the digitization of voluntary health insurance. Its software products analyze bills from healthcare providers and help identify signs that service charges may have been inflated.
Ingosstrakh plans to apply AI to other medical insurance processes as well. Similar solutions could also prove useful in other countries, where insurers face the same challenge of reviewing large volumes of documents and checking whether requested services comply with clinical guidelines.
