Check in
A short, accessible conversation about today. No copyrighted screening scale is reproduced in this prototype.
SERVICE AND SAFETY BOUNDARIES
Useful support without pretending to know more than we do.
DailyAble separates observable information, questions that still need human context, and actions a person can choose.
THE SERVICE LOOP
Explain
Show what changed in plain language without converting it into a diagnosis or opaque risk score.
Choose
Offer one manageable next step and let the participant decide whether to involve a supporter.
Follow through
Supporters see observable reasons, permission scope, and the agreed next review—not unrestricted health data.
RESPONSIBILITY MAP
Three layers. No blurred authority.
LayerMay doMust not do
RulesValidate inputs, apply consent, route explicit safety conditionsInvent a new clinical threshold
Language modelRewrite for clarity, organize summaries, suggest questionsDiagnose, prescribe, decide an emergency, or contact people autonomously
HumansChoose support, confirm context, make professional decisionsTreat the system output as a medical conclusion
DATA MINIMIZATION
If a field creates no help, action, or safety value, we do not need it.
Participant control
Choose the supporter, information scope, and when sharing ends. See who receives a notification before confirming it.
Purpose limits
No advertising segmentation, public labels, medical diagnosis, or model training from participant support data.
Prototype honesty
This deployment uses simulated records. In-memory demo actions reset when the service restarts.
TAIWAN CONTEXT
Why accessible daily support matters
of persons with disabilities surveyed in 2024 had never used the internet.
Ministry of Digital Affairs sourceadult diabetes prevalence reported in Taiwan’s 2017–2020 nutrition and health survey.
Health Promotion Administration sourcedementia prevalence among adults aged 65 and above in the 2020–2023 national community survey.
Ministry of Health and Welfare sourcePopulation statistics describe the public challenge. They are not used to infer an individual participant’s condition.