MedGPT: AI Use Reveals Primary Care Strain & Needs

Analysis of MedGPT.fr usage, published on March 17, 2026, by Synapse Medicine, reveals not only the rise of a medical conversational AI assistant, but as well provides insight into the realities of clinical doubt as healthcare professionals navigate decisions about patient care. The data highlights the increasing pressure on frontline providers, who are often required to respond quickly with limited resources.

Key Takeaways

– More than 65,000 healthcare professionals were already using MedGPT.fr six months after its launch.

– Usage is primarily focused on aiding diagnosis and medication prescription.

– General practitioners make up the majority of inquiries, but nurses and pharmacists also heavily utilize the tool.

– The nature of the questions suggests an immediate need for support within a strained healthcare system.

– This growth occurs as the framework for AI usage in healthcare is still being developed.

A Rare Snapshot of Needs Expressed During Patient Care

Launched in September 2025, MedGPT.fr is presented by Synapse Medicine as a conversational medical assistant designed for the French healthcare system. Six months later, the company reported that the tool was already being used by over 65,000 healthcare professionals. The report published on March 17, 2026, is based on statistical analysis of a representative sample of questions actually asked on the platform. Here’s significant because it’s based on observed usage, rather than self-reported data, providing a closer look at real-world situations.^[1]^[2]

This data sheds light on the fast-paced environment of clinical practice. When diagnostic hesitation arises, or when questions about indications, dosages, or treatment plans emerge, professionals first seek a readily available answer. According to the company, in 49.6% of cases, the expected response needs to be “clear and immediately actionable.”^[1]

The breakdown of professions using the tool confirms its integration into primary care. General practitioners represent 52.6% of the analyzed questions, followed by nurses at 15.1%, and pharmacists at 13.7%. Together, these groups account for over eight out of ten inquiries. The AI conversational tool isn’t being adopted for marginal or exploratory uses; it’s being integrated where demands are highest, time is limited, and clinical uncertainty is greatest.^[1]^[2]

This trend aligns with evolving professional landscapes, including the expansion of the nursing role to include consultation, diagnosis, and prescription and the rapid adoption of AI tools in medical practice, without eliminating human oversight.

Diagnosis and Prescription at the Core of Usage

The primary finding of the report centers on the nature of the questions asked. The tool isn’t primarily used for ongoing education or theoretical research. It’s mobilized in the core of clinical activity. Assistance with diagnosis accounts for 55.4% of inquiries. Within this category, professionals describe clinical symptoms to obtain guidance in 15.8% of cases, submit exam results for refined interpretation in 12.7% of cases, or seek a structured diagnostic approach for complex situations in 10.2% of cases.^[1]

Assistance with medication prescription is another major component of usage, with 47.5% of questions. Indications are the most frequent request, at 15.6%, while dosages account for 10.2% of inquiries. These figures demonstrate that MedGPT isn’t being used remotely from care, but rather in the midst of a therapeutic decision.^[1]

An infographic accompanying the report details the form of the requests. Factual questions dominate, at 33.2%, followed by yes/no questions at 16.4%, lists at 7.3%, summaries at 4.8%, and requests for sources or bibliography at 3.1%. This indicates that professionals are seeking targeted, structured, and immediately usable support rather than lengthy explanations.^[2]

Nurses and Pharmacists: A Strong Signal of Evolution in Primary Care

The breakdown of usage by profession is perhaps the most notable element of the report. Nurses dedicate 44.0% of their questions to diagnostic assistance, while pharmacists dedicate 27.8%. Conversely, questions related to prescription reach 36.5% among nurses and 67.9% among pharmacists. General practitioners utilize diagnostic assistance for 52.9% of their requests, compared to 33.7% for medication prescription.^[2]

These figures do not, on their own, indicate a breach of legal scope of practice. However, they document a need for cognitive support in situations where professionals are more directly exposed to triage, guidance, securing a prescription, or interpreting clinical signs. The report describes a system where the boundaries of roles are increasingly influenced by the realities of patient pathways rather than solely by regulations.

“What this analysis reveals goes far beyond the tool itself. It’s a window into what French healthcare professionals truly experience on a daily basis. The data shows that nurses and pharmacists are facing increasingly complex situations, often beyond their usual scope. MedGPT.fr is there to help them navigate these challenges with greater confidence and less isolation.”^[1]

This interpretation aligns with evolving trends in the professional field, including the official recognition of initial prescription rights for advanced practice nurses. It also reflects a broader shift in primary care characterized by longer wait times, a scarcity of medical resources, and a redistribution of demands placed on non-physician healthcare providers.

Rapid Growth, But a Framework Still Under Construction

This growth occurs during a period of ongoing regulatory development. In the fall of 2025, the French National Authority for Health (HAS) published its initial guidelines for the use of generative AI in healthcare. Then, on March 5, 2026, the HAS and the National Commission for Informatics and Liberty announced a public consultation on a joint guidance document. The signal is clear: adoption of the tools is progressing rapidly, but the framework for practice is still being developed.^[3]^[4]

this report should be carefully considered. It doesn’t provide a comprehensive clinical validation of the responses provided by the tool, nor does it assess the impact of those responses on the quality of decisions made. However, it does document with remarkable clarity the intensity of the needs expressed by healthcare professionals at the point of care. It suggests that if medical AI is finding its place so quickly, it may be less because of its allure and more because it’s filling a critical gap in already overburdened environments.