Why the alarm clock hasn't really changed in 100 years — and what we did about it

Mechanical alarm clocks were patented in the 1840s. Since then we've added LCDs, swapped buzzers for sampled music, moved the whole thing onto our phones, and bolted on sleep tracking. But the core loop is the same as it was in 1847: a noise goes off, you press a button to make the noise stop, and you decide whether to get up or do the whole thing again in nine minutes.

You know how that loop usually ends.

The problem isn't the noise

If you've tried any of the "smart" alarms over the years — math puzzles to dismiss the snooze, photo challenges that make you walk to the bathroom, QR codes you taped inside the fridge the night before, sleep-cycle detection that promises a "gentle" wake-up window — you already know the pattern. They work for about two weeks. Then your half-asleep brain gets good at them, and you're back to snoozing through everything.

The reason isn't that the obstacles aren't hard enough. It's that none of them actually engage you. They just gate the snooze button. Your brain treats them like a captcha: solve it on autopilot, fall back into bed, repeat.

What does reliably get people out of bed, in our experience, is something different: another person talking to you. A partner, a roommate, a parent yelling up the stairs. Not because they're loud. Because you have to respond, and you can't respond on autopilot. A conversation occupies the part of your brain that wants to be unconscious.

We built an alarm around that idea.

What ClockClanker actually does

When your alarm fires, you don't get a snooze button. You get a "Talk to ClockClanker" button. Tap it, and an AI voice starts talking to you in real time — by name. It tells you the day, your local weather, what's on your calendar, the thing you said last night you wanted to do first this morning. Then it asks you something, and waits for you to answer.

You answer out loud. It responds. The conversation continues until you're out of bed.

There's no script. Every morning is genuinely different — the AI is generating the conversation live, based on the actual context of your day. If you mumble "five more minutes," it pushes back. If you go silent, it nudges you — gently at first, then more directly, then urgently. There's no version of this where you can roll over and let it go.

You pick the voice (ten of them, five male and five female), the language (English, Spanish, Portuguese, German, or Italian), and the tone — Gentle if you want warmth, Firm if you want accountability. You set the things you actually want to hear about in the morning: a goal you're working on, an affirmation, the first concrete action you want to take when your feet hit the floor.

That's the whole product. One mechanic, executed carefully.

The boring part that matters: privacy

A talking AI alarm clock raises an obvious question: is it listening to me all night?

No. Here's the actual flow:

• Your microphone only opens after you tap "Talk to ClockClanker," when the alarm is already ringing.
• Voice activity detection and speech-to-text run on your iPhone, locally. Your audio never goes anywhere.
• Only the resulting text — what you said, converted to characters — is sent to our server.
• That text is forwarded to the AI model that generates the next thing ClockClanker should say. Our server doesn't store it. OpenAI keeps API requests briefly under their standard policy (up to 30 days for abuse monitoring, then deleted) and doesn't use it for training.
• The response comes back, and your phone synthesizes it into speech locally, on-device.

Your voice recordings, your GPS coordinates, the contents of your calendar at rest — none of that ever lives on our servers. We chose this architecture before writing a line of conversation code. It's also why the app requires a recent iPhone: the on-device speech model only ships on iOS 26 and up.

No ads. No data selling. No tracking. The whole company runs on the Pro subscription — you can manage or cancel it from your Apple ID at any time, and there's a free trial so you can see whether it works for you before paying anything.

A morning, end to end

The night before, you set an alarm for 7:00. You tell ClockClanker:

• Your first action tomorrow is "twenty pushups before checking my phone."
• Your affirmation is "I don't need a perfect day to have a good one."
• A morning topic: "the marathon I'm training for."

At 7:00, your phone rings. Through silent mode. Through Focus. Even if you force-quit the app last night — alarms are scheduled through Apple's AlarmKit, the same system the built-in Clock app uses.

You tap Talk to ClockClanker.

"Morning, Alex. Quarter past seven. Sixty-one and cloudy in Berlin, rain coming in around two — you'll want a jacket. Your calendar's clear until that 10:30 with the team. Before any of that: twenty pushups. You ready to get those done?"

You grunt. It pushes back. You eventually answer. The conversation continues for two or three minutes — sometimes longer if you let it go — bouncing between your day, your training, whatever's on your mind. By the time you've ended it, you're out of bed, you've moved your body, and you already know what's ahead of you.

That's the whole thing. It's not magic. It's just the obvious move nobody had taken seriously yet: replace the dismiss-button game with a conversation that actually wants something from you.

Try it

ClockClanker is on the App Store for iPhone (iOS 26 or later). There's a free trial, and the subscription is the only way we make money — no ads, no data sale, no upsells inside the app. If it doesn't change your mornings within a week or two, cancel it from your Apple ID settings and we'll never bother you again.

We'd love it if it did.

— The team at ClockClanker