QUINCY
Awareness for every body
01The awareness layer for care robotics

The world’s AI learned from everyone but you.

Built with disabled and neurodivergent people — not just for them.

Quincy is a hardware-flexible AI layer that helps care robots understand the speech, behavior, and patient needs conventional systems miss.

▶ Watch it work For partners →

Quincy started with you.

Built for you first — not an accessibility feature bolted on at the end, but the whole reason Quincy exists. And hardware-flexible, so it works on whatever robot is already in the room.

02The people machines were never taught to hear

Millions speak in ways
machines call noise.

Disability and neurodivergence. Stroke, injury, aging — and anyone a hurried system talks past. Permanent, temporary, or a season of recovery: real voices and bodies that AI trained on “average” leaves behind.

Quincy was built to change that.

03Hardware-flexible by design

So it runs on any body.

Quincy licenses into existing robots rather than replacing them — it’s the intelligence layer, not the chassis. Its most complete build runs on the MARS manipulator, with focused work on a Unitree G2 (medicine retrieval) and an SO-101 arm (reading with kids). Tomorrow, any care robot on the market.

Most complete on
Innate MARS
Also on
Unitree G2 · SO-101
Layer
Hardware-flexible
04Three systems, one intelligence

You’re heard. You’re seen. You’re not alone in it.

Voice · Beta
VOICE
It understands the way you actually speak — not the way a dataset expects.
Care · Beta
CARE
It notices when something’s wrong and checks in — on your terms. Nothing leaves without your say-so.
Advocate · In development
ADVOCATE
When the paperwork and the systems won’t budge, its agents push back — so you’re not doing it alone.
▶ See the agent negotiate
05Under the hood

Built, not imagined.

Real models running on real robots — with honest tags on what’s beta and what’s next.

CP-aware pain & mood model
Multi-modal — speech, vision, and context — built to read pain and mood in people with cerebral palsy. Best Multi-Modal AI, Innate Robohacks 2026.
Speech-pathology-aware STT
Speech-to-text that understands atypical speech, trained on Nebius (NVIDIA GPUs). In beta.
Embodied on three platforms
Most complete build on the Innate MARS manipulator, with focused demos on a Unitree G2 (medicine retrieval) and an SO-101 arm (reading with kids).
06Team
Alison Cossette
Occupational therapist turned AI founder

I spent years as a clinical occupational therapist in oncology and end-of-life care — with the people machines understand worst. I became a data scientist to help more patients than I could reach one at a time. Quincy is my full circle: everything the bedside taught me, built into the robots now standing beside it.

  • —Clinical occupational therapist. Oncology & end-of-life care — years at the bedside of the patients care robots serve.
  • —M.S. Data Science (AI), Northwestern University.
  • —Neo4j Developer Advocate. Graph-native systems thinking.
  • —5× hackathon winner building at the edge of embodied intelligence.
  • —Award-winning Physical AI. Best Multi-Modal AI, Innate Robohacks 2026.
Jeff Sirianni — COO

Employee #3 at Embody, scaled from startup through its acquisition by Zimmer Biomet for up to $275M.

07Build with us

Robots that get
the rest of us.

Bringing Quincy to the Frontier Tower residency — to build the AI that finally works for the people it was made for.

Get in touch →
For disabled & neurodivergent people·patients & families·care-robotics teams & partners