| --- |
| title: Parent Co-Pilot |
| emoji: 🧸 |
| colorFrom: purple |
| colorTo: indigo |
| sdk: gradio |
| sdk_version: 6.16.0 |
| python_version: '3.11' |
| app_file: app.py |
| pinned: false |
| license: mit |
| short_description: Tiny-model parenting support for the next messy stretch. |
| tags: |
| - track:backyard |
| - achievement:offbrand |
| - decision-support |
| - parenting |
| - tiny-titan |
| |
| --- |
| |
| # Parent Co-Pilot |
|
|
| Built by Clarissa Chen for the Build Small Hackathon. |
|
|
| Parent Co-Pilot is an AI-powered parenting support tool for moments when a parent needs a realistic plan quickly. It is not just an activity recommender: it uses household context, parent energy, child age, sickness, weather, space, budget, cleanup tolerance, and timing to generate a constraint-aware plan for the next stretch. |
|
|
| ### Demo Video |
| - [Watch Demo](https://youtu.be/S8tV1G5DAsI) |
|
|
| ### Blog Post |
| - [Read Blog Post](https://www.clarissachen.com/blog/parent-co-pilot-build-small-hackathon) |
|
|
| ### LinkedIn Post |
| - [Read LinkedIn Post](https://www.linkedin.com/posts/clarissafchen_ai-llm-huggingface-share-7472286081623797760-c2EI/?utm_source=share&utm_medium=member_desktop&rcm=ACoAABSVo2kBRrerVzK63w9_gwLtEnz2C6LqBVo) |
|
|
| |
| ## What It Does |
|
|
| Parents enter a household profile and the current situation, then Parent Co-Pilot returns: |
|
|
| - Situation assessment |
| - Recommended plan with a timeline |
| - Lower-effort alternative plan |
| - Emergency fallback for when the plan breaks down |
| - Per-plan tradeoffs for parent effort, interruption risk, and cleanup burden |
|
|
| The app handles practical scenarios such as sick children, sick parents, both parent and child being sick, solo parenting, too-hot/rainy days, work calls, limited space, low energy, and low-cleanup needs. |
|
|
|
|
| ## Tiny Model |
|
|
| The default model is: |
|
|
| `HuggingFaceTB/SmolLM2-135M-Instruct` |
|
|
| This keeps the app aligned with the hackathon's small-model spirit while staying responsive on Hugging Face Spaces CPU runtime. The model does the planning from the user's context; deterministic code is used only for guardrails, formatting, and fallback behavior if generation fails. |
|
|
| Runtime knobs: |
|
|
| - `MODEL_ID`: defaults to `HuggingFaceTB/SmolLM2-135M-Instruct` |
| - `MAX_NEW_TOKENS`: defaults to `420` |
| - `ACTIVITY_CONTEXT_LIMIT`: defaults to `10` |
|
|
| ## Gradio Setup |
|
|
| Parent Co-Pilot uses Gradio, but the interface is customized beyond the default component look. The app uses a one-column guided flow, custom CSS, icon badges, soft plan cards, timeline rows, and per-plan tradeoff chips to make the result feel like a practical parenting dashboard instead of a generic chatbot. |
|
|
|
|
| ## Local Run |
|
|
| ```bash |
| pip install -r requirements.txt |
| python app.py |
| ``` |
|
|
| Then open the local Gradio URL shown in the terminal. |
|
|
