The established GP IT market is facing pressure from both new core-system entrants and AI suppliers seeking a place in the clinical workflow. Doctolib's May 2026 acquisition of Medicus connects those pressures. Its planned NVIDIA minority investment, announced on 13 October 2026, adds a further statement of AI ambition, not evidence that the existing suppliers have already been displaced. Doctolib's acquisition announcement and NVIDIA announcement.
EMIS Web and TPP SystmOne have been deeply embedded in general practice. Changing that position requires more than a better demonstration: a new system must support the operational work on which practices and patients depend.
The Incumbent Position
The core GP system connects prescribing, referrals, appointments, clinical notes, QOF reporting, tasks and documents. Switching therefore involves more than teaching staff a new interface. Records, national services, local processes and ongoing clinical work all need to remain usable during and after migration.
The earlier article described migration in terms of millions of patient records and weeks of disruption. That scale may be relevant to a large migration programme, but it should not be presented as an inevitable duration or record count for every individual practice. The local scale, sequencing and contingency plan must be established rather than assumed.
Switching costs can help protect incumbents, but they do not prove that product quality is irrelevant or that switching is practically impossible. Familiarity, reliability, support and existing capabilities may also be rational reasons to remain. NHS England's June 2025 announcement nevertheless recognised the limited choice practices had historically faced and made greater competition an explicit objective. NHS England.
NHS England Wants Change
Medicus's approval through the Tech Innovation Framework in June 2025 was described as the first major new entry in a quarter of a century. NHS England also said at that time that more suppliers meeting the framework's standards were expected by April 2026. That was a dated forecast, not a future milestone that should still be described as upcoming in October 2026. NHS England's announcement.
The original coverage recorded more than 25 Integrated Care Boards exploring adoption pathways by mid-2025. Exploration is not completed procurement or deployment. Its significance is evidence of interest in alternatives, not proof that all those organisations had switched systems.
The distinction matters because policy intent, supplier assurance, purchasing decisions and successful implementation are different stages. A competitive market requires credible movement through each stage, not simply an increased number of announced products.
Medicus Proved the Gate Can Be Opened
Medicus's initial achievement was passing the relevant assurance requirements, not obtaining an official verdict that it was the best GP product. Its public account describes integration with 24 national NHS services and implementation of five new national open standards for information exchange. Those services include EPS, eRS, NHS App, PDS and SCR. Medicus.
Information governance, data protection and clinical safety, including the applicable manufacturer-side DCB0129 work, form part of the wider assurance discussion. Developing and testing those connections can require substantial engineering and clinical effort. There is not one universal number of years that every entrant must spend, and assurance of the core system is not blanket approval of every future AI function. NHS clinical safety standards.
The practical result is that a new entrant has demonstrated a route into the market. Doctolib has acquired that organisation and its accumulated experience, while still needing to deliver reliable implementations and support beyond the initial sites.
Doctolib Brings Capital, Product Experience, and European Scale
The May 2026 article recorded Doctolib's reported reach as 520,000 health professionals and 90 million people across France, Germany and Italy. Those are historical reach figures, not a current count of active AI users or evidence that every person received care through the same service. The earlier phrase Europe's largest platform should not be used as a clinical-quality comparison.
The financial dates also need to remain explicit. The reported EUR 348 million annual recurring revenue figure relates to 2024, as reported in May 2025, not to newly disclosed October 2026 revenue. The previously cited valuation exceeding EUR 6 billion is historical financing context, not a disclosed valuation for NVIDIA's planned investment. No direct comparison should be made with OpenEvidence's US-dollar figures by silently converting currencies or mixing transaction dates. Reporting on Doctolib's 2024 ARR.
The May 2026 plans for more than £100 million of UK investment, 150 hires and a London research and development centre indicate an intended long-term commitment. They are not a statement that the entire amount has already been spent. Doctolib's announcement.
Its earlier acquisitions also show several expansion routes: Dottori.it for Italian market entry, Siilo for professional messaging, Aaron.ai for German telephone assistance and Typeless for speech recognition. These are distinct transactions and capabilities, not proof that one identical acquisition playbook will work in every country. Dottori.it transaction advisers, Siilo announcement, Aaron.ai adviser and Typeless account.
The Real Prize: The Integrated Clinical Desktop
The strategic objective is broader than a booking system or a scribe. It is an environment in which appointments, consultations, documentation, prescribing, referrals, messaging and clinical decision support can work together. The operating-system analogy describes that ambition rather than a recognised regulatory or technical classification.
The October 2026 NVIDIA announcement adds AI development resources to this picture. Doctolib reports more than 12 million patient visits documented by its consultation assistant and identifies administration, coding, billing and clinical decision support as development areas. Those usage and roadmap statements do not establish that every function is deployed in Medicus or available for NHS procurement. Doctolib's announcement.
For a practice, daily reliability remains the decisive test. A system must preserve access to the record, support authorised actions, handle exceptions and provide a usable fallback. A prominent investor does not answer whether a particular practice can migrate safely or whether its staff can complete essential work during disruption.
Competition Can Change the Economics of Supplementary Tools
An independent AI supplier may previously have sold directly to individual clinicians. If core systems bundle similar capabilities, the supplier might need to demonstrate stronger specialisation, more transparent evidence or better implementation across several platforms.
Alternatively, a more open core-system market could create demand for modular services. Evidence retrieval, professional learning and specialist workflow functions might be purchased separately and connected where appropriate. That is a plausible market structure, not proof that every vendor will permit identical access.
The current GP Connect documentation, reviewed on 15 October 2026, illustrates why interoperability should be discussed by service and permission. A national interface is not a promise that every third-party application can read or write everything in every system.
A Fictional Practice's Migration Test
Imagine a fictional practice choosing between remaining on its existing system and moving to a new platform. The new interface is attractive, but the team identifies unresolved questions about historical attachments, outstanding tasks, staff training and fallback prescribing arrangements.
A responsible decision would address those issues before treating an AI roadmap as a reason to accelerate migration. The practice could test representative workflows, establish which records and tasks require reconciliation, define acceptance criteria and agree who can postpone the transition if essential work is not ready.
The same approach applies to staying with an incumbent. Familiarity should not prevent examination of avoidable friction, poor support or new alternatives. The comparison should be between realistic operating arrangements, not a flawless new platform and a caricature of an old one.
What This Means for Clinical AI Tools
This article is published by iatroX and includes its own reference and educational role. As described on 15 October 2026, iatroX offers guidance-grounded clinical answers, calculators, examination preparation and CPD tools independently of a single EHR vendor. This does not imply a live native integration with each core system. iatroX methodology and learning catalogue.
For a practice selecting an EHR, integration, reliability and migration support are central. For a clinician checking evidence or continuing professional learning, an independently accessible tool may preserve continuity when the workplace changes. The case for modularity is strongest when it improves the user's actual work, not merely because independence sounds desirable.
The GP IT market can become more competitive without every capability belonging to one platform. The clinical knowledge layer should remain inspectable and usable, whether embedded or independent.
Turn this reading into evidence you can use
A professional learning activity can focus on one concrete question, such as distinguishing core-system assurance from approval of an additional AI function. The clinician should review and personalise any draft reflection and suggested framework mapping rather than presenting generated text as learning that did not occur.
As described on 15 October 2026, iatroX CPD tools are included within the paid learning subscription and support reviewed learning records. That is separate from accreditation, regulatory assessment or proof of competence to manage an EHR migration. iatroX CPD.
Frequently asked questions
Has the GP IT duopoly already ended?
The market has gained a new approved entrant and additional competitive pressure, but an announcement or planned deployment is not evidence that incumbent systems have been displaced.
Should a practice change EHR because a supplier has stronger AI funding?
Funding is relevant context, but the decision should depend on the actual system, migration plan, support, integration and evidence for the functions the practice will use.
Does an independent learning tool replace core GP software?
No. Clinical reference and professional learning serve different purposes from maintaining the patient record and executing practice workflows.