Parenting. Without the chaos.

The technical centrepiece of the build. Parents upload their legal parenting agreements (custody schedules, visitation rules, holiday arrangements, court orders) directly into the app. icodelabs built a complete RAG pipeline using LangChain and Weaviate: documents are parsed, chunked, and embedded into Weaviate's vector database. When a parent asks a question — "Am I allowed to take the kids on holiday this weekend?" — the system retrieves the most relevant passages from their specific document and passes them as context to GPT-4, producing answers grounded in their actual legal agreement rather than generic AI responses. LangChain-Experimental was used for advanced document reasoning chains across complex multi-clause legal documents.

A conversational AI assistant was built into the React Native app, powered by GPT-4 via LangChain. The chatbot provides real-time co-parenting coaching, conflict resolution guidance, and communication strategies — all informed by the user's uploaded parenting plan context. NLP sentiment analysis detects passive-aggressiveness, emotional triggers, and manipulation patterns in messages from an ex-partner, and the AI suggests neutral, de-escalating replies calibrated to the specific situation.

A dedicated Python backend was built using FastAPI and Uvicorn to handle all AI inference, RAG pipeline orchestration, document processing, and LangChain workflow execution — separate from the Node.js application backend. This dual-backend architecture keeps AI processing isolated, scalable, and independently deployable, while the Node.js layer handles all standard application logic, authentication, and API routing.

A library of AI-generated, scenario-specific message templates was built for common co-parenting situations — schedule change requests, expense discussions, holiday coordination, and discipline decisions. Templates are generated by GPT-4, customisable by the parent, and sent directly via Twilio SMS or Email API integration — keeping communication structured, neutral, and legally defensible.

An emotional support module was built with AI-powered daily stress management tips, mindfulness reminders, and a "Calm Mode" — a one-tap feature that instantly surfaces AI-generated coping strategies for high-stress moments. Child-centric advice features provide AI guidance on how to talk to children about separation, handle anxiety, and frame difficult situations in age-appropriate language.

A subscription-based iframe tool was built for family lawyers to embed on their own websites. The widget provides AI-generated responses for handling high-conflict co-parenting enquiries — giving lawyers an intelligent first-response tool for prospective clients. The embeddable snippet is distributed through the Next.js web app with subscription-gated access for law firms.

OneSignal was integrated for targeted push notifications — delivering reminders for court dates, visitation schedules, holiday arrangements, and mindfulness prompts extracted from the user's parenting plan by the RAG pipeline. Notifications are personalised based on each user's document-specific schedule.

Ghost was integrated as a headless CMS for the Parent Co-Pilot blog and resource library — providing SEO-optimised content around co-parenting, legal guidance, and emotional wellness to drive organic acquisition.

Supabase handles structured relational data — user profiles, parenting plan metadata, subscription records, and message history. Firebase handles real-time features and authentication via Firebase Auth. Weaviate operates as the dedicated vector store for document embeddings, keeping the RAG pipeline's retrieval layer separate from application data.