What we went to BIG to understand

BIG, Bpifrance's annual entrepreneurship event, took place on October 8, 2026 at Accor Arena in Paris. The Healthcare team at Dedale Intelligence, Soukeyna Ndiaye and Carla Daubet, attended to follow the French HealthTech sessions: a panel on AI adoption and hospital transformation with Nabla, CHU Grenoble Alpes, Hublo and the Ministry of Health, and sessions with Doctolib, Medadom, Arlequin AI and Olva, a European MedTech start-up.

For investors, corporates and M&A advisors, a day like this offers something reports cannot: what buyers and vendors say when they share a stage. We compared what we heard with our published research, including our Healthcare Survey '25 and our work on AI in healthcare and electronic health records. Our wider perspective on the segment is in Dedale's take on HealthTech and our research on the digital transformation of healthcare IT. Five signals stood out.

Hospitals buy savings they can measure

Hospital and public-sector speakers were direct about how purchasing decisions are made: tools with a visible effect on costs win over tools that mainly improve productivity. Hublo, a workforce management platform present in about 6,000 healthcare structures, was cited as an example of a solution that can replace an administrative function, which makes its economic contribution easy to measure. AI documentation tools were described differently. They can give clinicians time back, but that time does not automatically translate into lower spending or additional revenue. Nabla, which sells such a tool, estimates that automating clinical reports, discharge letters and handovers can save about 15% of clinician time in relevant care settings.

Our research points the same way. In our Healthcare Survey '25, which covers 101 healthcare IT decision-makers, integration complexity, high upfront costs and limited ROI were the main barriers among organizations making limited use of emerging technology, and 36% of institutions said they prefer to wait for proven ROI. The French EHR and hospital information system market is itself small: we size it at about $400-500 million across roughly 1,330 hospitals, so new tools compete for limited budgets.

For vendors, a time-saving claim is not enough. A hospital buyer wants to see which line of the budget changes.

AI scribes face commoditization

Nabla expects standalone AI scribing to be commoditized within two to three years. To stay ahead, the company is expanding into specialty workflows such as cardiology, adjacent uses such as medical coding, end-to-end clinical documentation and offline capabilities. Hospital speakers added that an AI application adds limited value if it cannot integrate into clinical workflows and reliably transfer structured information into the patient record, and that overlapping functionality from incumbent software vendors weighs on selection.

Our landscape work explains why. Our report on AI in healthcare rates AI medical scribing as a high-value new use case, one notch below automated ICU documentation, which we rate very high. Our mapping of ambient AI in France shows Doctolib with newly launched native ambient functions alongside point solutions such as Nuance, Nabla and Heidi, while EHR vendors in the US and Germany build ambient AI in-house. A point solution that depends on integration into another vendor's platform faces a platform that can build the same function.

The two-to-three-year timeline is one vendor's expectation, so we treat it as a hypothesis. The direction matches what we see in the market: depth in specific workflows and integration into the patient record are becoming the test.

France trails the US on clinical AI, largely because of billing

Nabla puts France 24 to 36 months behind the United States in ambient AI adoption. The company entered the US market in 2023 and describes a market that has moved to replacing existing solutions, citing a deployment covering about 10,000 physicians at a Los Angeles children's hospital.

The reason it gave is financial. In the US, AI-generated clinical documentation can support medical coding and billing to private insurers, so time saved becomes revenue. In France, clinical reports are not the basis for billing and hospitals lack a dedicated budget for these tools. Nabla argues for outcome-based payment models to reduce purchasing risk for hospitals, and says international expansion is essential given the limited size of the French market.

Adoption itself is present in our data: 68% of hospitals in our Healthcare Survey '25 use or pilot automated clinical documentation tools. A pilot is not a paid, scaled deployment, and the gap Nabla describes may sit in that difference. The 24-36 month figure comes from a single vendor, so we regard it as a hypothesis to test with French hospital buyers.

Explore our research on the digital transformation of healthcare IT

Interoperability and financing still limit scale

Speakers returned repeatedly to interoperability. Continuity of care requires continuity of patient information across hospitals, outpatient providers and other care settings, but there is no consensus on common clinical metrics and data frameworks, so the problem goes beyond technical integration. Regional care pathway initiatives still rely heavily on manual processes, and speakers called for data platforms that can support coordination and serve as infrastructure for AI applications. One panel also raised a concern that incumbent software vendors may slow interoperability discussions while they define their own AI strategies, which creates uncertainty for hospitals and third-party vendors.

Our survey data shows how widespread the issue is. 59% of respondents cite interoperability as the most significant ongoing challenge, ahead of legacy user interfaces that require costly customization (51%) and limited influence over vendor roadmaps (41%). 38% connect 20 to 50 applications to their EHR or hospital information system, and 24% connect more than 100.

Financing is the other constraint. Public-sector speakers described the difficulty of funding AI from general hospital budgets without dedicated mechanisms, and noted that France's activity-based reimbursement continues to reward clinical activity over prevention or care optimization. In our research on the EHR market, public programs such as France's Ségur and Germany's KHZG are a main growth driver, in a market that grows under 5% a year. Spending follows dedicated funding, which suggests AI will need its own line in hospital budgets to scale. The concern about incumbents' AI strategies was raised in one panel, so we flag it as a point to test with buyers.

Platforms are expanding into content and new channels

Doctolib used its session to present a role beyond appointment booking. The company described prevention campaigns built with articles and videos co-created with healthcare professionals, with medical teams reviewing sources before publication, as a response to medical misinformation. It also described an AI assistant integrated into its platform, reportedly reaching about 10,000 users, with an initial focus on parents of young children. Panelists stressed the importance of curated medical knowledge bases for healthcare AI, given how easily generative models can surface unreliable information.

Medadom described a different route to scale. It uses pharmacies as a distribution channel for remote medicine, on the premise that everyone lives within 15 minutes of one. The company reports around 11 million teleconsultations and a device in about one in four pharmacies, and it is extending the model to specialty services in local retail: teleophthalmology in opticians' shops and teledermatology based on photos taken in the pharmacy. Hublo described a workforce platform that helps staff move between hospitals and medico-social settings.

This fits our view of Doctolib. Our report on the company shows 7 acquisitions that shifted from buying scale to buying capabilities, including AI call handling and voice recognition, around a patient engagement base that represents 60 to 70% of revenue (FY24 revenue of €273.7 million, up 19%). A patient-facing assistant and validated content extend that pattern toward prevention. All user and consultation figures here are company statements, and a 10,000-user assistant is an early signal.

Beyond software: MedTech scale and AI reliability

Two other sessions are worth noting. In a session on European MedTech with Olva, a medical device start-up that began as a spin-off from a French university hospital institute, speakers argued that Europe's challenge lies more in commercialization and market fragmentation than in innovation capacity. FDA approval opens a large and relatively unified US market, while CE marking does not remove country-level differences in reimbursement, purchasing and commercialization across Europe. Combined with a limited domestic market and insufficient venture capital at scale, this makes it harder for French MedTech companies to show traction and finance industrialization. For investors, commercialization capability and reimbursement access matter as much as the technology.

Arlequin AI, a Paris-based start-up founded in 2024, presented an alternative AI architecture designed to make model outputs traceable to their underlying evidence, arguing that models that are correct most of the time are not enough where errors can disrupt essential services. The discussion did not establish whether the approach performs better than existing architectures, so it remains a signal to monitor. Reliability and auditability could become adoption criteria for healthcare AI beyond generative capabilities themselves.

What this means for investors, corporates and M&A advisors

• Vendors of standalone AI tools face growing pressure to show measurable savings, integrate into the patient record and, where possible, connect to billing.
• Platforms with an installed base, such as Doctolib, and models built on workforce or channel access, such as Hublo and Medadom, have clearer routes to scale in France.
• Interoperability and dedicated AI funding remain the gating factors for hospital adoption, and both depend on public policy as much as on vendors.
• These are early signals from a single day of sessions, based largely on what speakers said. We will track which of them show up in buyer data and vendor results over the coming months.

Request access to our healthcare research on the Dedale Intelligence platform.

Contact us to discuss healthcare software with our team.