Helix

Helix came out of a hackathon. Me and David Adedeji decided to tackle the healthcare problem we kept hearing about from clinicians across Nigeria. Doctors are drowning in paperwork. The patient in front of them gets maybe a third of their attention because the rest is going into writing up notes, pulling charts, scheduling follow ups, and chasing down records from the last visit. Existing EMR systems were built for Western hospitals with dedicated admin staff. In most African clinics, the doctor is also the admin.

So we built Helix as an AI native EMR. One where a doctor can just say "pull everything on Mrs. Adebayo's last three visits and summarise what I need to know before she walks in" and get a clinician friendly answer in seconds. The agent is a two stage system. Gemini 2.5 Flash plans which tools to call, LangChain orchestrates the execution with structured output parsing, and the result comes back as a clean summary prioritising clinical findings over admin noise. We designed it with two faces. Doctors get a multi tab workspace and calendar, patients get a simple profile and AI booking chat. The whole thing is optimised for the reality that most clinic staff don't have time to learn complicated software. If it doesn't work in the first 30 seconds, they go back to paper.