How Kyiv-Based Bookimed Is Testing AI in Medical Travel

Bookimed, a medical-travel platform founded in Kyiv in 2014 with more than 1,700 clinics and 5,800 doctors across 32 countries, is automating how patients search for clinics and contact providers abroad. The company was established by Ievgen Khotianov and Yevhenii Kozlov. Its service connects prospective patients with clinics and doctors in different countries, while coordinators assist with requests, treatment options and communication with providers. JOIN US ON TELEGRAMFollow our coverage of the war on the @Kyivpost_official. Bookimed initially operated through a relatively simple model: patients submitted requests, after which coordinators helped them compare clinics and obtain information about treatment. The company continued operating after Russia’s full-scale invasion of Ukraine, with its team shifting to remote work. Automating first contact Bookimed has since introduced an AI coordinator, Sofia, which handles first contact in 14 languages around the clock. An early version was built within days and outperformed the conversational software used before in internal tests, by the company’s own measurements rather than an independent assessment. Sofia logs a request, searches the database, suggests clinics or doctors, prepares options and answers follow-ups; it also contacts clinics on a patient’s behalf. Bookimed says it does not diagnose or make clinical decisions, and cases needing medical interpretation go to human coordinators. The distinction matters: a platform’s suggestions can influence where patients seek treatment, even when framed as administrative rather than medical guidance. Oversight and data limits Medical information the service publishes is reviewed by a doctor, the company says, and a second system checks responses for unsupported claims. That addresses a known risk of generative AI, credible-looking but inaccurate output, without removing the need for human review, particularly on procedures, eligibility or expected outcomes. Output also depends on the database: inaccurate or outdated clinic data affects recommendations regardless of the model. Kozlov says keeping it accessible and structured is now a main operational priority. Changes in coordination Patients who hold longer conversations are likelier to continue than those sending one message, Bookimed reports, reading this as a sign that detailed replies and follow-up affect whether an inquiry proceeds. Engagement is not evidence of better outcomes or provider choice: the figures describe communication and conversion inside the platform, not care quality. How Bookimed earns revenue Bookimed does not charge patients: it takes a commission from a clinic when a patient arranges treatment and, it says, adds no markup. The model is common among intermediaries, though it can create potential conflicts between commercial interest and neutral comparison; its usefulness depends partly on whether relevant providers appear regardless of commercial ties. AI does not remove that question, it shifts part of the recommendation to systems shaped by the database, ranking criteria and business model. Agents developing agents Automated systems also review conversations, spot recurring problems and prepare software changes. According to Kozlov, groups of agents examine past interactions, draft adjustments, test them in simulations and release some after checks; the architecture is revised often rather than in rare updates. Faster testing raises the weight of oversight, documentation and rollback. The case shows AI agents used not only as customer-facing chatbots but as internal development tools, though internal claims do not show whether such systems would work elsewhere. For Bookimed, the immediate effect has been a change in how patient requests are processed and how software updates are developed. Whether the model improves medical-travel decisions depends on factors beyond automation: data accuracy, transparency and continued professional involvement. This article is published as part of Kyiv Post Spotlight and is not part of Kyiv Post’s editorial coverage.