We didn't pivot into AI. We started inside it. Every system we ship — from smartphone dermatology to ontology platforms to quantum research — is designed by people who treat truth as an engineering requirement, not a marketing word.
When we watched the first wave of consumer AI products charm the world with plausible fiction, we knew we didn't want to add another voice to that chorus. The industry had fallen in love with fluency; we fell in love with accountability.
We started Deeproot Minds to build AI systems that know what they know — and more importantly, know what they don't. That meant rebuilding the architecture from the ground up: ontology-first, graph-grounded, provenance-aware, clinician-supervised, human-overseeable. Not retrofitted safety. Not post-hoc alignment. Rooted truth from the first line of code.
Our three offerings — smartphone HealthTech, social-robot mental health, enterprise ontology platforms — are different surfaces of the same system. Each is an answer to the question: what does AI look like when its first commitment is to the person on the other side of it?
We also invest quietly and seriously in quantum research, because the ontologies our clients will need in the 2030s won't fit on classical hardware. Building for that future isn't hype — it's homework.
Intelligence without roots is performance. We're here to grow the roots.
A system that says "I don't know" is more valuable than one that confidently invents. Every Deeproot product defaults to abstention over fabrication.
Healthcare, mental health and enterprise decisions don't reward move-fast-and-break-things. We rigorously test, calibrate, and validate before we ship.
Every claim our systems make must trace to a verifiable source — a guideline, a graph entry, a clinician's training data. No floating outputs.
We optimise for the patient, the child, the clinician, the operator — not for leaderboards. Our success metric is outcomes, not scores.
We publish. We collaborate with universities. We contribute to open ontologies. Our research group works adjacent to the product group — not downstream of it — so that the long-horizon questions shape the short-horizon deliverables.
Current focus areas include graph-native retrieval-augmented generation, multi-modal affect recognition in pediatric contexts, and variational quantum circuits for ontology embedding. If any of these resonate with your lab's work, we want to talk.
Contributing back to OBO Foundry and W3C semantic-web community standards.
RCT collaborations in planning with NHS England trust partners for 2026–2027.
Research affiliations across UK and European universities in AI, ontology and quantum.
Independent oversight committee reviewing every clinical and pediatric deployment.