Most alarms are a sound you silence. Fifi is a call you answer. It rings through Silent mode and a Focus — built on iOS 26 AlarmKit, so the alarm needs no network. Your Lock Screen shows an incoming call with a face and a name. Answer, and one of ten characters reads you your day: last night's note, your forecast, your headlines. An ASMR whisper, a film-noir detective, a nature doc, a drill sergeant, a pantomime baroness, a fairy godmother. 39 ringtones. A backup call. A streak.
Hey Product Hunt 👋
Every alarm app I've used treats waking up as a fight. The category leader
literally markets itself as the world's most annoying alarm clock.
But the real problem isn't the alarm — it's the ten seconds after. You
silence it, you're still horizontal, and you start scrolling. The first
moment of the day goes to nothing.
I wanted the opposite. The idea landed when I thought about how I actually
get woken up most mornings: by my cat. So Fifi doesn't beep. She calls, and
when you answer there's someone on the line who has your day.
Two things I learned building it:
**Reliability stopped being a moat.** iOS has never really let third-party
apps set alarms — before iOS 26 everyone kept a silent audio session running
all night, which dies on reboot or when you swipe the app away. That's the
cause of every "it didn't go off" review in this category. AlarmKit fixed it,
and it fixed it for my competitors too. Which meant the product had to be the
caller, not the clock.
**Characters have to differ by what they DO, not by adjective.** I sampled a
competitor's ten voices and seven were doing the same thing — "you're wasting
your potential, get up" — in different vocabulary. Mine are separated by
speech act: soothes, narrates, observes, orders, taunts, bestows. No two
alike. It's also the only test I found for whether a character still has
anything to say on morning 365.
I also cut CallKit, which gives you Apple's real incoming-call screen and
looked perfect. It needs a VoIP push from a server, so the alarm dies when
the phone is offline. An alarm that needs a network connection is not an alarm.
Happy to answer anything. Which character would you want calling you?
Report
@yancietroy I like that each character has its own personality. That should keep the calls from getting boring after a while.
Report
Maker
@oliver_graf1 That was one of the main ideas behind Fifi. The characters aren’t just different voices as each one has their own personality and tone, so the morning briefing feels different depending on who you choose.
There’s a pretty wide range too, from calm/ASMR-style characters to more energetic personalities like a sergeant who basically tells you to get out of bed.
The goal is to make the experience feel more personal than hearing the same generic alarm sound every morning. :)
Report
Hi there! Great idea for getting up everyday with smile, as you said an alarm that dies while you are offline is not an alarm, but if it was possible to have the CallKit design exactly the same as apple would you use it, ore you will still keep your to maintain you unique design of AlarmKit?
Report
Maker
@matteorobella Thank you! For Fifi, I’d actually keep the AlarmKit experience. The “call” is more of a fun gimmick rather than the core functionality as underneath, it’s still a reliable iPhone alarm.
Fifi uses Apple’s AlarmKit, so the wake-up screen behaves like the standard Apple alarm rather than trying to recreate CallKit. The character interaction happens after you dismiss the alarm, which is where we add the personality.
I also like the idea of using a telephone-style alarm tone to sell the call concept without sacrificing the reliability of a proper alarm.
The reframe from an alarm you silence to a call you answer is the part that stuck with me. I build voice AI at Callie Care, a companion that calls aging parents every day, and the biggest lever on whether someone picks up turned out to be the identity on the lock screen rather than the sound itself. A named face gets answered, an unknown number gets ignored. Question on the ten characters: do you see answer rates differ by persona, or does the choice mostly drive retention after the first week? On our side people pick a voice on day one and almost never switch, so that first pick carries a lot of weight. Also curious how you time the backup call.
Report
Maker
@igorgurovich Yeah, that’s an interesting comparison. With Fifi, the “call” is really a gimmick layered on top of the normal iPhone alarm experience rather than an actual phone call.
Fifi still uses Apple’s AlarmKit, so it behaves like a normal alarm and shows the standard Apple alarm screen. After you dismiss/answer the alarm, the character greets you and delivers the morning briefing.
You can also choose a telephone-style alarm tone to make the whole “Fifi is calling you” concept feel more convincing, but there’s no actual incoming call or caller ID involved.
Because of that, I’m not really measuring “answer rates” by persona in the same way you are. I’m more interested in whether the character/voice someone chooses makes the wake-up experience more enjoyable and affects whether they keep using Fifi.
The point about people rarely switching voices after the first choice is interesting, though. I could see the initial character selection having a similar effect for Fifi, especially because the character is what you interact with immediately after waking up.
@yancietroy I like that each character has its own personality. That should keep the calls from getting boring after a while.
@oliver_graf1 That was one of the main ideas behind Fifi. The characters aren’t just different voices as each one has their own personality and tone, so the morning briefing feels different depending on who you choose.
There’s a pretty wide range too, from calm/ASMR-style characters to more energetic personalities like a sergeant who basically tells you to get out of bed.
The goal is to make the experience feel more personal than hearing the same generic alarm sound every morning. :)
Hi there!
Great idea for getting up everyday with smile, as you said an alarm that dies while you are offline is not an alarm, but if it was possible to have the CallKit design exactly the same as apple would you use it, ore you will still keep your to maintain you unique design of AlarmKit?
@matteorobella Thank you! For Fifi, I’d actually keep the AlarmKit experience. The “call” is more of a fun gimmick rather than the core functionality as underneath, it’s still a reliable iPhone alarm.
Fifi uses Apple’s AlarmKit, so the wake-up screen behaves like the standard Apple alarm rather than trying to recreate CallKit. The character interaction happens after you dismiss the alarm, which is where we add the personality.
I also like the idea of using a telephone-style alarm tone to sell the call concept without sacrificing the reliability of a proper alarm.
Awesome idea. Any plans for an Android version?
@eyalron definitely! will launch on Android soon
@yancietroy cool. will be happy to test it
Refocus
The reframe from an alarm you silence to a call you answer is the part that stuck with me. I build voice AI at Callie Care, a companion that calls aging parents every day, and the biggest lever on whether someone picks up turned out to be the identity on the lock screen rather than the sound itself. A named face gets answered, an unknown number gets ignored. Question on the ten characters: do you see answer rates differ by persona, or does the choice mostly drive retention after the first week? On our side people pick a voice on day one and almost never switch, so that first pick carries a lot of weight. Also curious how you time the backup call.
@igorgurovich Yeah, that’s an interesting comparison. With Fifi, the “call” is really a gimmick layered on top of the normal iPhone alarm experience rather than an actual phone call.
Fifi still uses Apple’s AlarmKit, so it behaves like a normal alarm and shows the standard Apple alarm screen. After you dismiss/answer the alarm, the character greets you and delivers the morning briefing.
You can also choose a telephone-style alarm tone to make the whole “Fifi is calling you” concept feel more convincing, but there’s no actual incoming call or caller ID involved.
Because of that, I’m not really measuring “answer rates” by persona in the same way you are. I’m more interested in whether the character/voice someone chooses makes the wake-up experience more enjoyable and affects whether they keep using Fifi.
The point about people rarely switching voices after the first choice is interesting, though. I could see the initial character selection having a similar effect for Fifi, especially because the character is what you interact with immediately after waking up.