Contact
An older adult in an armchair resting a hand on Guardian, a soft knitted companion robot with a heartbeat line on its visor

Parxyz The future of AI companions

Care that feels like family. Companions that check in every day, notice how you feel, and keep your loved ones and care team close.

Explore
Our vision

AI companions will soon live alongside us, at home, on our wrists and in our care. We're building ones that keep your data private, explain themselves, and know when to bring in a human.

38Publications
450+Citations
3Companions in development
2026Founded
Products

Three companions.
One trustworthy mind.

Each companion builds on our team's research in explainable, privacy-preserving and human-centred AI. Together they are designed to share what they learn, with your permission, across the home, the wrist and the care team.

Companion robot

Guardian

A home companion that checks in, talks things through and keeps you safe.

Prototype Close-up of Guardian's knitted head and glossy visor with a heartbeat line

Checks in

Daily conversations, reminders and gentle nudges, at a pace that suits you.

Watches out

Notices the early signs of a fall and checks before alerting anyone, so alarms stay meaningful.

Learns privately

Learning happens on the device, so personal data stays in the household.

Made for home

For older adults living independently, their families and home-care providers.

Wellbeing companion

Steady

A companion robot that reads the sensors around you, notices how you feel, and helps you and your clinician make sense of it.

Research prototype Steady, a chubby knitted companion robot with a calm wave on its face, holding up a small sensor while rings, patches and bands orbit around it

Sensors all around

Rings, patches, bands and home sensors track heart signals around the clock.

Talks it through

Gentle check-ins in plain language when your patterns change.

Open to challenge

Every insight shows its reasoning, and clinicians can correct it.

Clinician-connected

Shares summaries with your care team only when you choose.

Care companion

Chorus

An AI companion for the whole circle of care: older adults, families and care teams.

Research prototype A knitted house with warmly lit windows and a small Guardian robot by the door

A voice for everyone

Agents represent the priorities of each person involved in care.

Clear reasoning

Suggestions come with reasons anyone can follow and question.

Human sign-off

Nothing is adopted until the older adult and care team agree.

For care teams

Home-care agencies, long-term care providers and community health teams.

Images are concept renders. Parxyz products are in research and development. They are not medical devices, have not been approved or cleared by any regulator, and are not intended to diagnose, treat or prevent any condition.

Approach

How it works From signal to companion.

All three companions share one foundation: privacy-preserving learning on the device, and AI that explains itself and accepts correction.

  1. Sense

    Wearables, home robots and everyday conversation give the companion context about health and daily life.

  2. Learn privately

    Models run on the device and improve through federated learning, so personal data stays at home.

  3. Explain and contest

    The companion shows its reasoning, and users, carers and clinicians can question and correct it.

  4. Hand over to people

    When something matters, it brings in the right person, and a human always makes the final call.

Privacy by design

Edge-first processing and federated learning keep what leaves a home to a minimum.

People stay in charge

Companions suggest; people decide. Every output can be reviewed, challenged and overridden.

Honest by default

Companions explain what they noticed and why, building on our team's explainable-AI research.

Evidence before claims

We publish our methods in peer-reviewed venues before we promise outcomes.

Research

Research 38 publications

Our companions build on peer-reviewed work by the Parxyz team at venues including IJCAI, ACM Multimedia, Information Fusion, ACML and ICMR.

All publications Google Scholar
Team

The team building the future of AI companions.

Parxyz brings together computer science, biomedical engineering and health research. Our members are affiliated with the University of New Brunswick in Canada and Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) in Germany.

Truong Thanh Hung Nguyen smiling on a winter day above the frozen St. Lawrence River in Québec City

Truong Thanh Hung Nguyen

Co-founder & CEO

Hung co-founded Parxyz to bring trustworthy, human-centred AI companions out of the lab and into homes and clinics. Hung is a PhD candidate in Computer Science at the University of New Brunswick's Analytics Everywhere Lab, working on explainable and contestable AI, self-evolving multi-agent systems, edge computing and biomedical engineering.

Before Parxyz, Hung earned an M.Sc. in Artificial Intelligence (minor in Biomedical Engineering) from FAU Erlangen-Nürnberg and a B.Sc. in Computer Science from Frankfurt University of Applied Sciences, both as a DAAD scholar, and worked as an AI Scientist and Business Unit Director at FPT Software.

Email Google Scholar ORCID LinkedIn Website
Vo Thanh Khang Nguyen

Vo Thanh Khang Nguyen

Co-founder

Khang is an AI researcher working on explainable AI and reinforcement learning. Khang's research covers how to evaluate explanation methods, explainable object detection, and trustworthy AI for medical imaging such as thyroid nodule diagnosis, and has been cited more than 300 times.

Google Scholar LinkedIn
Hai Son Lam

Hai Son Lam

Co-founder

Hai Son works on artificial intelligence and law: how AI systems can reason with legal rules, and how law and regulation should govern AI. Hai Son's interests include AI governance, data protection and accountability for automated decisions. At Parxyz, Hai Son focuses on the legal and regulatory side of our companions, from privacy and consent to accountability when AI supports care decisions.

Hoang-Loc Cao

Hoang-Loc Cao

Technical Manager

Hoang-Loc works on trustworthy and human-centred AI, multi-agent systems and biomedical image analysis. Hoang-Loc is a research assistant at the University of New Brunswick's Analytics Everywhere Lab and holds a B.Sc. in Information Technology (Data Science) from the University of Science, VNU-HCM. Recent results include first place in the BIC-MAC Challenge at MICCAI 2026 and the Multimedia Verification Challenge at ACM ICMR 2026.

Google Scholar LinkedIn Website

Our members' academic affiliations

Contact

Work with Parxyz

contact@parxyz.sarl Write to us

Clinical partnerships, pilots, research collaborations and investment. Founder: hungntt@parxyz.sarl