This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

What I Built

I kept ending days with a messy head.

Notes half-written, tasks still floating, a vague feeling that something was unfinished. I would close the laptop and still feel the day following me into the evening.

I wanted something that would do the closing work for me - not another productivity system, just a quiet ritual that says “you can stop carrying this now.”

SlothCloser is that ritual.

You dump whatever is on your mind (messy is fine).

It gives you back:

  1. A short, human day summary
  2. The few things still open (so your brain can stop looping)
  3. A soft bedtime reflection shaped by how the day actually felt
  4. One tiny thing for tomorrow - optional, light, never a to-do list
  5. An ambient soundscape prompt you can imagine or paste into any free generator

Then it shows the only line that matters:

Day closed.

You can stop carrying it now.

That is the whole product.

It is also a small experiment in personal super-intelligence that stays local: a private evening companion that knows only what you tell it tonight, runs on an open-weight model you control, and never phones home.

Demo

Live: https://sloth-closer.onrender.com

(Free tier - first visit after idle can take 30–60 s to wake)

Local (recommended for real use):

ollama pull gemma2:2b streamlit run app.py

The public demo always works. If no local model is attached it falls back to carefully written offline responses so the ritual never breaks.

Code

GitHub logo meenanoolu / SlothCloser

A calm, local evening wind-down agent that helps you close the day - slowly, kindly, and completely offline.

🦥 SlothCloser

A calm, local evening wind-down agent that helps you close the day - slowly, kindly, and completely offline.

Built for the Hacktoberfest Open-Source AI Challenge: Week 3 – Slow and Steady

Live demo: https://sloth-closer.onrender.com

Why I built this

I kept ending days with a messy head.
Notes half-written, tasks still floating, a vague feeling that something was unfinished. I would close the laptop and still feel the day following me.

I wanted something that would do the closing work for me — not another productivity system, just a quiet ritual that says “you can stop carrying this now.”

SlothCloser is that ritual.

What it does

You dump whatever is on your mind (messy is fine).
It gives you back:

  1. A short, human day summary
  2. The few things still open (so your brain can stop looping)
  3. A soft bedtime reflection shaped by how the day actually felt
  4. One tiny…
View on GitHub

How I Built It

Core stack

  • Streamlit for a calm, focused UI
  • Ollama + open-weight models (Gemma 2 by default)
  • Pure Python, no cloud LLM required

Design principles

  • Tone is non-negotiable: warm, unhurried, slightly wise. Never corporate. Never a productivity coach.
  • Every prompt is short and permission-giving.
  • The final screen must produce the emotional payoff: “Day closed.”
  • Offline-first. The public demo degrades gracefully instead of failing.

Architecture

  • agent.py - tries the local model first, falls back to high-quality offline templates
  • prompts.py - all tone rules live here so the voice stays consistent
  • app.py - custom CSS dark theme, mood selection, session history, downloadable closing

The model is treated as a quiet friend, not a productivity engine. Temperature is kept low. Token limits are tight so small models stay fast and calm.

Why Does Open Innovation Matter?

This tool only makes sense with open-weight models.

Closed API Open-weight (local)
Your private evening notes leave the machine Notes never leave your laptop
Requires internet and ongoing cost Works offline after one download
Model is fixed and opaque Swap Gemma / Llama / Qwen / Phi any time
Feels wrong for something this personal Feels like a private ritual

A hosted chatbot would feel invasive for the exact use case this app is built for. Open models let the tool stay private, permanent, and free.

The Lazy Payoff

I have used it every night this week.

The real win is not the summary. It is the feeling that the day has been received and set down. Open loops stop looping. The bedtime reflection gives permission to stop. The final screen is deliberately short and quiet so the brain can actually release.

That is the laziness the challenge asked for: the work of closing the day is done for you, so you can rest.

Prize Categories

I am entering:

  • Best Use of Gemma - primary open-weight model powering the entire ritual
  • Best Use of Render - live demo hosted via Render Blueprint + Docker
  • Overall Challenge Winner

Built with ❤️ and deliberate slowness during Hacktoberfest 2026.

Take it slow. The day will still be there tomorrow.