Incoming: eye centre hotline, 2:14 am, 中文
Example caller, 74, sudden curtain over vision
One hospital team for a safe test. Nothing reaches a patient until you've seen it work.
我右眼突然看不清,好像有黑色的帘子挡住
"My right eye suddenly went blurry, like a black curtain"
It's 2:14 am. She's 74. Who picks up?
Where the phone queue goes wrong today
First come, first servedUrgent callers wait behind routine ones
Long waits on holdSudden vision loss can sit behind reschedules and follow-ups
Nurses on the phonesPhone work eats into patient time
What your department gets
Urgent callers first
Every call rated, so the sickest isn't fourth in line
Mdm TanBehind 4 routine calls today
With FirstheardEscalated first, under your protocol
Example
Red flags caught first
Fixed rules check red flags before the AI
Caller says"Like got a black curtain covering it"
Red-flag ruleMatched, before any AI
RouteWarm transfer to the on-call nurse
Example
Nurses back with patients
Phones and bookings handled, so shifts go to care
TodayPhones and bookings take part of every shift
With FirstheardThat time goes back to patients
Older callers served in their language
Built for English, Mandarin and Malay callers; dialects planned
CallerCalls the number she already knows
No appNothing to read, type or download
Your protocol decides
Unsure calls go to a person with the transcript
UrgentRoute now, with a summary
UnsureStaff take over
Your staffFinal say
Phone queue today: first come, first served
Reschedule
Follow-up
Reschedule
Question
Mdm TanUrgent: escalated
On hold:behind 4 routine calls
Waiting:music on hold, eye getting worse
Answered:first ring, in her own words
Urgent:escalated to the on-call nurse
Example queue
She waits behind four routine calls.
Your ED sees the sickest first. Firstheard is built to do that on the phone.
Singapore: eye care demand is growing
21.4%of citizens are 65 or olderPopulation in Brief, June 2026
426k+eye patient visits in 2024 at SNEC and its affiliated centresSNEC
+48%glaucoma in older Singaporeans by 2040, 57,800 to 85,800Gupta et al., IOVS 2025
I can't read the app, I just want to talk to someone
Example caller
App:small text, typing, passwords
Phone:the same number she already knows
Languages:built for English, Mandarin and Malay; dialects planned
Nurse shift: phones and bookings taking time from patients
Example shift: not measured data
Mdm Tan's call, with Firstheard answering
Every call rated, routed by your protocol
Urgent:route now, with a summary
Priority:slot in your urgent window
Routine:booked, no staff time needed
Unsure:staff take over with the transcript
Tomorrow 9 am: your eye department
Tomorrow at 9 am: overnight calls already sorted.
2:14 am:Mdm Tan escalated under your protocol
Overnight:routine calls already booked
9 am:nurses start with patients
Start with past calls and nothing reaches a patient
Past callsDe-identified, under an MOAStop here
Silent shadowingFirstheard listens, staff decideStop here
Supervised pilotOnly after you've seen the resultsStop here
An earlier version was built and tested with a design-partner private eye-clinic chain. The current system is an upgraded version.
Who should we speak to first?