Founder Memo

Healthcare is entering an era where we can decouple the capital invested from a project’s success: a single, well-equipped player can now build what once required an entire team and substantial fundraising.

I can think of no field more worthy than healthcare in which to innovate and build. Like many others, it will be profoundly transformed by AI — yet it will remain, for a long time still, carried by humans to be executed.

And so I believe the agentic approach must sit at the heart of a new generation of healthcare software, at every layer of how it works:

Karukera is a company I currently own individually and in full, and it will be the foundation for building an ecosystem centered on agentic, AI-first healthcare software.

AI, LLMs, agentic coding, and AI-assisted humans are at the center of building this tool, whose aim is to use SOTA processes in structuring and development to maximize performance, cost, the relationship with the user, and ultimately the functionality of the products built on it.

SuperPagr — whose name is bound to change, so unpronounceable it is — is the first step on this path, demonstrating what a single human, aided by these technologies, can achieve on a very precise business problem: the medical scheduling of continuity of care, with its challenges of user experience, algorithmics, and data consistency, in an environment of permanent technological lag.

The French hospital software market already has suites that try to cover everything — Maincare, Softway Medical, Dedalus foremost — but none carries the spirit that made an Odoo succeed: real modularity, transparent pricing, fast deployment, an ecosystem open to whoever wants to build on top. That space is vacant, and it is so for precise reasons, not by chance: HDS certification, interoperability standards (HL7, FHIR), medical liability, long public-procurement cycles, regulatory fragmentation. These very constraints are what make the problem hard to solve — and therefore, once solved correctly, hard to copy.

OpenStats, which tackles the statistical analysis of medical theses, is the second demonstration of this same method: a precise and painful business problem, poorly served by existing software, solved by a single human leaning on agentic AI. SuperPagr and OpenStats are not two isolated bets: they are the first two bricks of one same, repeatable approach to the healthcare market.

Imagine an ecosystem where the EHR and the software attached to it can answer us, draw up chemograms in real time, suggest novel therapeutic approaches, explain the latest report to the patient in plain language, or automatically collect and include lab results in the record.

Imagine a medical-data format, designed AI-first, compatible with any tool — letting the GP query the specialist’s latest treatment, the research lab request the extraction and analysis of a patient’s data, a nurse know the medical directives, and above all an LLM find its way within one of the most important sets of data for a human being.

Imagine software whose reach is obviously individual, but also collective: a standard we build and maintain, yet open to everyone’s use — in the manner of Stripe, proprietary in its governance but become, through the quality of its experience and its adoption, the de facto interface of an entire sector — and a proven, secure plumbing for agents, letting anyone come and build the tool they need, with SOTA data and standards respected and maintained.

Karukera is the name of Guadeloupe, a French land where ideas can be born, but often amid permanent chaos. The Seijaku approach — calm in the storm — embodies our desire to nurture a sudden, soothing calm within a system that cares, even at the heart of suffering, and to ease, at last, the transition to the new world.

This is only the beginning.