In the critical minutes after a stroke, the standard neurological exam is a human one. A paramedic or triage nurse asks a patient to smile, raise both arms, and repeat a simple phrase, looking for the telltale signs of facial droop, arm drift, or slurred speech. It is a high-stakes, time-pressured assessment, and its accuracy varies. A French startup, AI-Stroke, is betting that a 30-second smartphone video analyzed by artificial intelligence can do it better, and more consistently, for every patient who needs it.
Founded in 2022, the Montpellier-based company has just secured a $4.6 million seed round to advance what it calls an "AI neurologist" for pre-hospital and early in-hospital stroke triage [PRNewswire, Nov 2025]. The capital is earmarked for clinical validation and navigating the regulatory pathway toward FDA clearance in the United States, a necessary step for any software aspiring to become a standard tool in emergency medicine [Yahoo Finance, April 2025].
The clinical wedge is a video dataset
The company's technical foundation is a proprietary dataset, which it describes as the world's largest clinically annotated collection of stroke video. This repository contains over 23,000 videos and 4.5 million images of patients performing the standardized limb movements and speech tests used in protocols like FAST (Face, Arm, Speech, Time) [LinkedIn, retrieved 2026]. This dataset is the fuel for the AI model that paramedics would activate with a smartphone's camera.
The intended workflow is straightforward. A first responder records a brief video of the patient. The AI analyzes facial symmetry, arm movement, and speech patterns, returning an analysis within seconds [Neuronews International, April 2025]. The goal is not to replace the clinician's judgment but to augment it with a consistent, data-driven second opinion at the point of first medical contact, long before a CT scanner is in reach.
Early validation against human performance
AI-Stroke's most compelling claim to date comes from a study conducted as a training module with French emergency medical services (SDIS). Involving more than 2,000 first responders, the company reports its model detected twice as many 'true' stroke cases as the EMS personnel and outperformed human results across all key diagnostic parameters [Neuronews International, April 2025]. While full, peer-reviewed publication of these results is awaited, such early data provides a tangible signal of potential clinical utility.
The company's leadership brings a mix of technical and entrepreneurial experience. CEO Cédric Javault, a former division chief for France's Nuclear Safety Authority, leads the company with over 15 years of business leadership experience [AI-Stroke, retrieved 2026]. He is joined by co-founder and CBO Simon Jiafeng Li, a deeptech entrepreneur and investor who previously founded medtech diagnostics company Gensor, and CTO Fabien Orsat, who manages both technology and quality assurance [Crunchbase, retrieved 2026] [mines-paris.org, retrieved 2026].
Navigating a crowded and regulated field
The ambition to improve stroke care with AI is not new, and AI-Stroke enters a field with established, well-funded players. Competitors like Viz.ai, RapidAI, and Brainomix have gained traction and regulatory clearances, but their focus has largely been on analyzing medical imaging,CT and MRI scans,after a patient reaches the hospital. AI-Stroke's distinct wedge is its pre-imaging, smartphone-based approach, targeting the diagnostic gap in the ambulance or at the emergency room door.
| Company | Primary Focus | Key Regulatory Milestone |
|---|---|---|
| AI-Stroke | Pre-hospital/early in-hospital triage via smartphone video | Pursuing FDA clearance [PRNewswire, Nov 2025] |
| Viz.ai | AI-powered care coordination & image analysis (CT/MRI) | Multiple FDA 510(k) clearances |
| RapidAI | Imaging analytics for stroke, aneurysms, and pulmonary embolism | FDA 510(k) cleared products |
| Brainomix | AI-powered imaging biomarkers for stroke and lung disease | CE Mark and FDA 510(k) clearances |
The risks here are well-defined and significant. The regulatory pathway for a Software as a Medical Device (SaMD) that provides diagnostic support is rigorous, requiring robust, multi-site clinical studies to prove safety and efficacy. The $4.6 million seed round, while substantial, is likely just the beginning of the capital required to fund those trials and build a commercial operation capable of selling into complex, budget-constrained healthcare systems.
The next twelve months: trials and trails
The immediate roadmap is clear. The seed funding will be deployed toward multi-center clinical studies in both Europe and the United States, which are the essential groundwork for regulatory submissions [Yahoo Finance, April 2025]. Success in these trials is the non-negotiable milestone that will determine whether the technology transitions from a promising prototype to a reimbursable tool. Concurrently, the company must begin the arduous process of integrating its software into the existing technology stacks and workflows of ambulance services and hospital emergency departments.
For patients suffering an acute ischemic stroke, time lost is brain lost. The current standard of care in the pre-hospital setting relies on the trained eye of a paramedic using screening scales, followed by a race to a comprehensive stroke center for definitive imaging. This process, while life-saving, has inherent variability. AI-Stroke is betting that an AI, trained on tens of thousands of clinical videos, can reduce that uncertainty, helping to ensure that every patient showing subtle signs of a stroke gets on the fastest possible path to treatment. The next year of clinical data will show if that bet can hold up under regulatory scrutiny.
Sources
- [PRNewswire, Nov 2025] AI-Stroke Raises $4.6M Seed Round to Advance AI-Powered Stroke Triaging Before CT Scan | https://www.prnewswire.com/news-releases/ai-stroke-raises-4-6m-seed-round-to-advance-ai-powered-stroke-triaging-before-ct-scan-302617124.html
- [Yahoo Finance, April 2025] Medtech Startup AI-Stroke Raises $4.6 Million to Bring Artificial Intelligence to Pre-CT Emergency Stroke Care | https://finance.yahoo.com/news/medtech-startup-ai-stroke-raises-090700665.html
- [Neuronews International, April 2025] AI-Stroke Raises US$4.6 Million Seed Round to Advance Pre-CT Stroke Triage | https://neuronewsinternational.com/ai-stroke-raises-us4-6-million-seed-round-to-advance-pre-ct-stroke-triage/
- [AI-Stroke, retrieved 2026] Home | AI-Stroke | https://www.ai-stroke.com/
- [LinkedIn, retrieved 2026] AI-Stroke | https://www.linkedin.com/company/ai-stroke/
- [Crunchbase, retrieved 2026] Simon Jiafeng Li profile | https://www.crunchbase.com/person/simon-jiafeng-li
- [mines-paris.org, retrieved 2026] Fabien Orsat profile | https://www.mines-paris.org/
- [BFMTV, July 2023] Cédric Javault, fondateur et directeur général d'AI-Stroke | https://www.bfmtv.com/economie/replay-emissions/tech-and-co/cedric-javault-fondateur-et-directeur-general-d-ai-stroke-17-07_EN-202307170719.html