MINTAG – Always Ready for School

Auf Deutsch lesen

Every morning, my kid used to stand at the door wondering: Do I need my gym shoes? Did I pack the math book? Will I freeze if I wear just a hoodie? Nine times out of ten, something was missing. So I decided to build a little tool to handle the worrying for them.

Enter MINTAG: a script that sends my kid an email each evening with tomorrow’s schedule, the weather, and even a hint about what to wear. It’s small, but it makes school mornings calmer.

The idea

I wanted something automatic. No new apps to install, no clicking around. Just an email that’s already there by the time homework is finished. Simple, reliable, invisible.

Under the hood

The workflow is straightforward:

• A Stundenplan file with the weekly lessons and required materials.
• The SRF Weather API to know tomorrow’s Swiss forecast.
• Apertus, the open Swiss LLM — specifically mlx-community/Apertus-8B-Instruct-2509-4bit — to generate a friendly daily message.
• An SMTP client that delivers the email safely to their inbox.
• A scheduler that wakes up the laptop at 18:00, runs the task, and then lets it go back to sleep.

Runs on a MacBook Air

Everything runs locally on a MacBook Air M4, 16 GB RAM. The laptop wakes from sleep, does the whole job, sends the email, and goes back to sleep — all in under three minutes. The heavy lifting (Apertus inference plus email composition) takes less than a minute. The rest is just wake-up, work, sleep.

What worked

Apertus was the most impressive piece. Instead of just listing “Math, Gym, bring notebook,” it writes in a way that feels friendly and encouraging. The messages are consistent, useful, and even a little creative — exactly the kind of reminder a kid actually enjoys reading.

Takeaway

MINTAG isn’t a polished product. But for my kid, it turned mornings from stressful to smooth. And for me, it was a reminder that even small, local LLMs can solve real everyday problems when pointed at the right spot.

Apertus and SRF Weather API are independent brands and are in no way affiliated with this project. The weather data used comes from the SRF Meteo API of SRG SSR. All rights reserved. This project is a personal initiative.