Born in the clinic.
Built for the patient.

We knew something was missing.

A medical history can stretch across years, different specialties, investigations and treatments that haven't helped. Sometimes the most useful clue is an old report, a seemingly minor symptom or a question nobody has had time to ask.

But consultations are short. Patients cannot always recall every detail on the spot. And even when clinicians recognize that a case deserves a deeper investigation, systematically exploring it is difficult in the middle of a busy day.

We first built tools for ourselves.

For more than a decade, we had been developing medical software. As AI models grew more capable, we built internal clinical decision-support tools to organize evidence, examine difficult questions and explore possibilities across specialties. Their potential was encouraging—but they required sustained attention.

Some of our most promising tools were also the hardest to use precisely when we needed them. The technology wasn't the only challenge. The clinical workflow was.

What if the patient could do the preparation?

Instead of asking doctors to fit another sophisticated tool into a packed clinic, we imagined a system patients and caregivers could use at home, at their own pace.

They could return with an organized history, findings worth revisiting and a clearer set of questions. The clinician would still examine, judge and decide—but the conversation could start from a better-informed place.

The patient should not need medical expertise to benefit from a medically informed investigation.

Access to AI wasn't enough.

Many patients were already searching Google or speaking to general-purpose AI tools. Yet the information they brought to the clinic was often difficult to use. We realized that the quality of a medical AI conversation depends on knowing which questions to ask, what context matters and which assumptions to challenge.

That expertise should not be a prerequisite for the patient. It should be built into the process.

A guided investigation—not a replacement for care.

WhatDx brings frontier AI models into a structured medical analysis process. It asks targeted questions, reviews symptoms and previous reports, explores possible explanations across specialties and helps identify useful next steps to discuss with a doctor.

Patients can work through the questions without having medical knowledge themselves. Clinicians retain responsibility for diagnosis, examination and treatment. Our aim is to make the time they spend together more useful.