Heylth AI Get Early Access

Health AIOS Eliminate Friction in Healthcare Access.

Heylth AI is building the next generation Workforce Health AIOS, bridging personal context with insurer- and employer-sponsored benefits. It recognizes when a need emerges and brings forward the support already available—so healthcare can proactively find you.

iOS Native Context → Needs
A private, native surfacebrings personal data together,turning signals into context.
Proactive Needs → Skills
Context reveals what matters,Doni recognizes the need,before friction slows action.
Frictionless Skills → Action
Move from a clear need to care—without portals or searching,without fragmented handoffs.
9:41
Good Morning!
TODAY Your context, together.
7h 12mSleep 42 msHRV 3Meetings
Doni

I noticed your sleep has shifted. Your plan includes support that may help. Want me to prepare the easiest next step?

Next step preparedCovered support · Tuesday 6:30 PM

Nothing is booked without your permission.

ADVICE STOPS BEFORE ACTION.

Where most health AI stops
01 Data 02 Diagnosis 03 Advice 04 Action
Execution is the product

The next healthcare platform is not another answer engine. It carries intent across eligibility, benefits, coordination and booking—until advice becomes completed care.

01 Context A vague signal:
“Something feels off.”
Cognitive friction What does this feeling mean?

People must translate ambiguous discomfort into a need clear enough for an AI or care system to understand.

02 Needs A need the system
can understand.
Fragmented friction Where can this need be solved?

Scattered vendors, coverage rules and unfamiliar GUIs make the right capability difficult to find and use.

03 Skills A capability that
can actually help.
GUI friction How do I actually finish?

Eligibility, approvals, booking, claims and reimbursement add another chain of work before care happens.

04 Action Care is approved,
booked and completed.
Advice availableAction completed

Health Cost Burden, Low Follow-Through.

Coverage already exists 92%

Americans insured in 2024

53.8% employment-based
Employer holds the exposure 67%

covered workers self-funded

All covered workers67%
200+ employee firms80%
Pressure keeps rising 9.2%

2025 cost trend before plan changes

Medical & pharmacy$4,798 lost productivity*
Natural follow-through 2030%

Natural-state adherence

Adherence distributionAdherence drives intervention effect
Available20–30%Outcome
Execution effect Follow-through compounds health outcomes.
Sources Census KFF Aon IBI

High Spend, Low Utilization.

Funded demand coverage Coverage Already Exists
92%

of Americans had health insurance for some or all of 2024. Coverage is not the same as completion.

Coverage already exists
More supply Hundreds

of benefits, vendors, accounts, apps and portals can sit inside a single employer ecosystem.

Friction reappears in the last mile Where utilization gets lost
01Recognize a fuzzy signal 02Translate it into a need 03Find the relevant benefit 04Verify eligibility and coverage 05Book, claim or reimburse Employee drop-off
What employers get back
Benefit UtilizationLow
Employee SatisfactionLow
$584 Benefit administration cost per employee, per year

Figures reflect the current Heylth market thesis and employer-benefits research; final publication sources should be confirmed before launch.

Access improved, Friction remains
01Human consulting era

Brokers

Expanded employer benefit portfolios,negotiated access and packaged vendorsupply into purchasable programs.

UnlockedSupply & Procurement
Friction left: More programs became available,but employees still had to know what existedand recognize when each benefit applied.
02PC internet era

Portals

Centralized plans, vendors and links,turning a fragmented benefit cataloginto one searchable destination.

UnlockedAggregation & Discovery
Friction left: Discovery improved, but employeesstill had to learn an unfamiliar GUI and choosethe right entry point for each need.
03Mobile internet era

Navigators

Added guidance and routing acrossan increasingly complex benefit stack,making options easier to compare.

UnlockedNavigation & Guidance
Friction left: Employees still initiated the journey,repeated context at every vendor handoffand carried the task through to care.
04AI era

Chatbots

Converted expressed questions to answersand matched stated intent with relevantbenefits more quickly.

UnlockedNeed Matching
Friction left: Matching improved, but eligibility,permission, booking and follow-throughstill remained the employee’s work.
SupplyAggregationNavigationMatching
The missing primitive: execution.

Heylth closes the loop: context becomes a need, a usable skill and a completed action.

The Chinese University of Hong Kong
Carnegie Mellon University
Dartmouth
Duke University
Hong Kong University of Science and Technology
The University of Hong Kong
Northwestern University
Stanford University
University College London
University of Pennsylvania
University of Washington
Massachusetts Institute of Technology
University of the Arts London
Technical University of Berlin
Henry, Heylth co-founder
CO-FOUNDERHenry, PhDUniversity College London · Health Informatics
Frayer, Heylth co-founder
CO-FOUNDERFrayer, PhDDartmouth College · Engineering
Elva, chief scientist
CHIEF SCIENTISTElva, PhDDuke University · Chemical Pathology
Tasman, software engineer
SOFTWARE ENGINEERTasman, MScMassachusetts Institute of Technology · AI
Galen, algorithm engineer
ALGORITHM ENGINEERGalen, MScCarnegie Mellon University · Computer Science
Leo, software engineer
SOFTWARE ENGINEERLeo, MScMassachusetts Institute of Technology · EECS
Cecilia, designer
DESIGNERCecilia, MScTechnical University of Berlin · Design
Catherina, product manager
PRODUCT MANAGERCatherina, PhDYale University · Public Health
Amazon
Apple
Oura Ring
WHOOP
GARMIN
Claude
DeepSeek
Gemini
Google
Microsoft
NVIDIA
OpenAI
Samsung
Slack
Snapchat
Spotify
WhatsApp
Fitbit
Doni Downloads

Meet Doni in Beta.

Done