LucidQuery Partners with Diagnoza to Power Its Health Companion AI
LucidQuery has signed a two-year agreement to power Diagnoza, the personal health companion for anxious minds and chronic bodies. Here is why a product built on reassurance needs an AI built on truthfulness.
A partnership built on trust
Today we are announcing a partnership with Diagnoza, the personal health companion for anxious minds and chronic bodies. Under a two-year agreement signed this month, LucidQuery now powers the AI behind Diagnoza's health and diagnostic-support stack. Every conversation a Diagnoza user has, every symptom they log, and every word of reassurance they receive is now driven by LucidQuery models.
We do not take that lightly, because of where it runs. This is not a casual chatbot. It is a product people lean on during flare days and anxious nights.
What Diagnoza is
Diagnoza puts it simply: "Gentle support, always on." It is a companion for people living with chronic conditions, health anxiety, and the long adjustment that follows a diagnosis, along with the people who care for them. Its job is to remember and to reassure:
- Persistent, encrypted memory of every conversation and symptom, so a user never has to start over or re-explain their history.
- Daily logging of mood, pain, and medication, building a timeline rather than a snapshot.
- Pattern monitoring that helps anticipate flares or panic episodes before they peak, with calm, breathing-led support in the moment.
- Reassurance grounded in evidence-based medicine, in plain language, with no jargon.
- Clinician-ready summaries and exports, so people walk into appointments prepared and their doctor sees an organized history.
Diagnoza is also clear about what it is not. In their own words, "Diagnoza is a companion, not a doctor." It does not replace clinical care. It remembers, organizes, reassures, and helps people arrive prepared, then leaves the medical decisions to the professionals who make them.
Why this is one of the hardest places to put an AI
In most products, a confident wrong answer is a bad experience. In a health companion for anxious and chronically ill people, it can do real harm. A model that invents a drug interaction, fabricates a statistic, or breezily dismisses a symptom that actually warrants attention is not just unhelpful; it can frighten someone who is already frightened, or quiet a worry that should have been escalated.
This is exactly the failure mode that ordinary language models are prone to. Left alone, they produce fluent, confident text whether or not they actually know the answer, because sounding right and being right are not the same objective. In this domain, that gap is unacceptable.
Why Diagnoza chose LucidQuery
Our models are built around a single discipline: be truthful, ground every claim that can be grounded, and refuse to fabricate. When something can be checked against real medical evidence, we check it. When something cannot be verified, the correct response is to say so plainly and point the person toward a clinician, not to manufacture a confident guess. We would rather give a careful, narrow answer than a broad and reassuring one that happens to be wrong.
That posture is the same one behind AGI-01 and the World Engine: ground the answer, verify it, and hold the line on what is actually known. It is what keeps our hallucination rate low, and it is precisely the property a health companion needs in order to be safe to lean on. Diagnoza chose LucidQuery because the values their product is built on, honesty, restraint, and care, are the values our models are built on too.
What this means in practice
LucidQuery now powers the conversational intelligence across Diagnoza: maintaining continuity over months of personal history, reflecting symptoms back accurately over time, producing clinician-ready summaries without inventing detail, and, most importantly, knowing the limits of its role. The model is built to never diagnose, never contradict clinical advice, and surface emergency resources when a situation calls for them.
Two years, and the bar that comes with it
A two-year commitment means we are accountable for how this behaves over time, across thousands of ongoing, deeply personal conversations. We are proud that a team this careful about the people it serves trusts our models to hold that standard, every day, on every flare day and every anxious night.
If you are building in a domain where being wrong is not an option, this is the bar we hold ourselves to.