Our mission
Advancing how humans and intelligent machines live and work harmoniously together
What we're proving first
No all-in-one promises.
We prove a few high-frequency, low-risk tasks first — cold.
Conversation and activity
Natural company through the day, with gentle nudges toward daily routines.
Reminders that matter
Non-medical-grade nudges for routines, medication, and the things that can't be missed.
Daily-living support
Hands-on help that takes load off families and caregivers.
Care coordination
Logs what changes and, with authorization, keeps the care team in sync.
Why now
A leap in AI capability
Models can finally reason in general — not just run one trick.
Falling hardware costs
Robot hardware and supply chains are finally cheap enough to deploy at scale.
Structural demand
Aging populations and a shrinking care workforce create demand that won't go away.
Not one breakthrough — models, hardware,
and demand maturing at once.
A window.
Why start with care
Care is one of the most important frontiers for physical AI.
It is emotionally rich, operationally complex, and deeply human. It demands systems that can perceive context, understand individual needs, act within clear boundaries, and improve through real-world feedback. This is where embodied AI becomes more than a demo. It becomes a relationship.
System architecture
What sets us apart
We sit where four hard things meet.
AI cognition
LegacyFixed scripts, passive responses
CyanRootGeneral understanding, proactive action
Robotic systems
LegacySingle-function devices
CyanRootSystems that act in the world
Human-machine relationship
LegacyOne-off tool use
CyanRootA relationship that grows
Real care settings
LegacyGeneric, far from the field
CyanRootBuilt on the care floor
We don't start from a single robot function, we start from the real tasks inside a long-term care relationship.
We don't aim for one-off interactions, we aim for continuous understanding, reliable assistance, and care coordination.
Where we are now
We're early. We're building the first care-companionship prototype and validating it in small, real-world steps. We don't claim what we haven't shipped — everything here has already happened.
Trust and safety
Capabilities are still growing. The principles aren't up for grabs.
These are design commitments, not finished features.
Explicit authorization
We process information only within an explicit authorization, and only as far as needed.
Recorded and reviewable
Key actions and state changes can be logged, replayed, and reviewed.
Human takeover
Caregivers and on-site leads can step in at any time.
Stoppable anytime
Users and caregivers always keep the right to stop the system.
Health and reminder features are non-medical-grade.
To make intelligence work in the real world
Partnerships · hi@cyanroot.com|Careers · talent@cyanroot.com