School Parent Copilot

base workflow

Elevator pitch

Ever feel overwhelmed by endless school emails? From trip reminders to newsletters, managing your child's school communications can be exhausting. Meet your new AI assistant that automatically sorts, categorizes, and summarizes school emails, turning chaos into actionable insights - all while running locally for privacy and sustainability.

Inspiration

As a parent, I was drowning in school emails. Trip notifications, newsletters, event updates, attendance reminders - each requiring different actions. The endless stream of communications made it impossible to stay organized and ensure nothing important slipped through the cracks. I needed an intelligent solution that could understand context and prioritize what actually mattered.

What it does

School Parent Copilot is a self-hosted AI workflow that transforms how parents manage school communications:

Smart Email Processing: Using n8n automation, the system monitors Gmail for incoming school emails and automatically routes them to an AI agent for analysis and categorization.

Intelligent Classification: The AI analyzes email content, extracting key information like deadlines, required actions, and importance levels, then categorizes emails into meaningful groups (events, newsletters, urgent notices, etc.).

Actionable Outputs: After processing, the system:

Saves structured data to a local database
Sends concise summaries via Home Assistant notifications
Could create calendar events and reminders automatically
Could add items to appropriate todo lists
Could route notifications to specific family members or devices

Smart Filtering: The system intelligently skips routine notifications (like "your child boarded the bus") that don't require parental action, focusing only on emails that matter.

All prompts are managed through Langfuse for consistent AI performance and easy optimization.

How we built it

Infrastructure Stack:

Docker: Containerized deployment for easy management and portability
n8n: Local workflow automation engine using the local-ai-packaged distribution, which includes Ollama for running local AI models
Langfuse: Prompt management and AI model performance tracking
Home Assistant: Smart home integration for notifications and automation
Local Database: Secure data storage for processed email information

Workflow Architecture:

Gmail API integration monitors incoming emails
School email detection and routing
AI-powered content analysis and categorization
Structured data extraction and storage
Multi-channel notification delivery
Action item generation and family coordination

Challenges we ran into

Hardware Limitations: Running multiple AI models locally on an NVIDIA RTX 2060 while maintaining other applications required careful resource management and optimization. The smaller model variants had to be used to ensure smooth operation. (I originally wanted to try this model on a Raspberry Pi 5 but I had the one with only 8gb for a 12Gb model - it would have a been a struggle anyway)

Email Variability: School communications come in wildly different formats, from formal notices to casual updates. Training the AI to consistently extract relevant information across this variety required extensive prompt engineering and testing.

Integration Complexity: Coordinating between n8n workflows, AI processing, database storage, and Home Assistant notifications while maintaining reliability and performance was more complex than initially anticipated. A lot of the nodes are there to provide flexibility (when a node is removed or added, the workflow is less likely to need rework!)

Accomplishments that we're proud of

Privacy-First Architecture: Built a solution capable of being completely self-hosted to keep sensitive family and school communications on local infrastructure, ensuring data privacy and security.

Environmental Sustainability: When powered by solar energy, the system operates as a carbon-neutral AI workload, demonstrating how green technology can enhance daily life.

Modular Extensibility: Created a flexible framework that can easily be extended with new features, adapted for different schools, or modified for other family communication challenges.

Real-World Impact: Successfully reduced email overwhelm from a daily stress to automated background processing, freeing up mental bandwidth for what really matters - being present with family.

What we learned

Local AI is Viable: Complex, intelligent workflows can be successfully implemented on consumer hardware with careful optimization and the right toolchain.

Automation Impact: Even automating seemingly small tasks like email categorization can have a disproportionately large impact on daily stress and productivity.

Integration Power: The combination of n8n, local AI models, and smart home systems creates powerful possibilities for practical AI applications beyond traditional use cases.

What's next for School Parent Copilot

Multi-Platform Support: Expanding beyond email to handle school app notifications, as many schools are moving away from email communication to dedicated platforms.

WhatsApp Integration: Adding support for parent group chats with intelligent monitoring for urgent messages while filtering out noise.

Scalability Options: Exploring cloud deployment options for families without technical expertise or suitable hardware, while maintaining the privacy-focused local option.

AI Model Optimization: Upgrading to more powerful hardware and larger models as the system proves its value and usage patterns become clearer.

Community Features: Open-sourcing core components to help other parents build similar solutions and share prompt engineering improvements.