AI & Med Tech

OpenAI has introduced ChatGPT for Clinicians, an artificial intelligence tool designed for verified US healthcare professionals and available free of charge. Built to fit into clinical workflows, the service brings together three core functions within a secure account: medical information research, documentation support, and the ability to earn continuing medical education credits. Its purpose is to give clinicians access, at the point of care, to answers grounded in millions of trusted medical sources, with citations that include details such as article titles, journals, and publication dates. ChatGPT for Clinicians also offers higher usage limits for GPT-5.4, which OpenAI presents as its most capable model for healthcare. It can assist with the analysis of complex cases, the comparison of treatment options, the synthesis of guidelines or recent studies, and the preparation of documents for review. The examples provided include the outpatient management of suspected community-acquired pneumonia, migraine prevention, and the assessment of several treatments while taking contraindications and monitoring requirements into account. The service can also ask questions to clarify the context of a case study before summarizing the evidence and explaining why the selected sources are reliable. Other suggested uses include drafting a referral letter to a specialist, creating discharge instructions in plain language, and producing multilingual versions once the original has been approved. Privacy is central to the offering: conversations held within the secure account are not used to train the models. OpenAI also emphasizes that the tool supports clinical reasoning and writing without replacing the healthcare professional. Clinicians therefore retain control over care decisions, while artificial intelligence serves as an aid for finding, checking, organizing, and documenting medical information more efficiently.

Every year, more than 400 million surgical procedures are performed worldwide, and an increasing proportion of them are recorded. Yet once an operation is over, this footage remains largely untapped. Uncovr aims to transform this body of clinical material into structured data. The startup has announced a $7 million funding round, approximately €6 million, led by Index Ventures alongside Seedcamp, Frst, No Label Ventures, Entrepreneurs First and several specialist investors. Already deployed in operating rooms across the United States and Europe, its platform analyses surgical footage second by second to identify the actions performed, their sequence, the events that occurred and the decisions made during the procedure. It then automatically generates a structured operative report, together with the information required for administrative coding. The implications extend far beyond saving time. According to data cited by the company, most of the operative reports examined omit more than 70% of the recommended clinical information. Early deployments also identified significant or billable steps missing from the documentation in 16% of cases, with an average reimbursement gap of around 10%. These omissions can affect continuity of care, the analysis of complications and the improvement of surgical practice. Uncovr’s real value, however, may lie in the gradual creation of a vast database of structured surgical data. For investors, surgery now represents a considerable reservoir of largely untapped data, comparable to medical imaging a decade ago. After patient records, imaging and the automation of medical documentation, artificial intelligence is now beginning to enter the surgical procedure itself. Automated report generation is the first practical application, while the data collected could power future surgical assistance systems and give rise to a new infrastructure capable of making surgery measurable, analysable and usable by artificial intelligence.

With thousands of scientific papers and medical guidelines published every day, Vera Health aims to give clinicians immediate access to medical knowledge. Founded by French entrepreneurs Taieb Bennani and Maxime Allouch, this specialist search engine scans more than 60 million research papers and drug monographs in real time. Its purpose is not to generate medical answers autonomously, but to identify, summarize and rephrase evidence already available in the literature. Every result is sourced and assigned a confidence level, and the tool can choose not to answer when the evidence is insufficient. Vera also tailors its responses to each clinical setting: an emergency physician receives concise information within seconds, while an oncologist can request a more comprehensive report. Several specialist models work in parallel to classify the question and adjust the format of the response. The startup reports strong performance on the NEJM-AI and MedXpertQA benchmarks, as well as on the US medical licensing examination, with its output continuously reviewed by a committee of clinicians. Adoption is accelerating rapidly. In France, the number of questions handled each month rose from 7,000 in January to 200,000 in June 2026, while more than 24,000 physicians are said to have adopted the application within six months. Internationally, Vera claims more than 400,000 clinician users. Selected by Y Combinator, the company has also partnered with the American College of Emergency Physicians to integrate guidelines for its 40,000 members. Following a $3 million funding round led by Gradient, Google’s artificial intelligence investment fund, the company is now expanding its service with personalized “Spaces.” Physicians and healthcare institutions can add their own documents, allowing them to combine the latest global scientific literature with their department’s internal protocols.

With OPTIMABIO, artificial intelligence is moving into the heart of an everyday yet highly consequential clinical task: ordering laboratory tests. Led by AP-HM, Hospices Civils de Lyon, Limoges University Hospital and French startup Kiro, the project has secured more than €17 million through the France 2030 iDémo programme. Its ambition is to analyse clinical and laboratory data in real time in order to recommend the most relevant tests, identify examinations that have already been performed and flag potential inconsistencies with medical guidelines. The final decision remains firmly in the clinician’s hands. OPTIMABIO also stands out for being integrated directly into existing hospital software. Rather than becoming yet another application, bringing additional interfaces and disrupting established workflows, AI is embedded as a native capability within the hospital information system. This approach could improve the quality of care while reducing unnecessary prescriptions, which consume resources, slow patient management and may trigger cascades of further testing. The economic stakes are substantial: according to the OECD figure cited in the document, around 20% of healthcare expenditure is linked to inappropriate care. The project also reshapes the role of university hospitals. No longer merely testing grounds, they become co-producers of the algorithm through their protocols, guidelines and clinical expertise. Beyond laboratory medicine, the same model could extend to imaging, treatment choices, patient referral and discharge planning. Challenges remain, particularly around interoperability, data governance, explainability and the long-term business model. Nevertheless, OPTIMABIO points to a profound shift: less spectacular AI, perhaps, but AI embedded in the thousands of routine decisions that shape every day in a hospital.