Back
Tech & Technique
Next.js, PostgreSQL, OpenAI, tRPC, TypeScript, Supabase, PostGIS, Retell AI, Twilio, Tailwind CSS, n8n
Description
A senior living guidance platform that matches older adults and their families with personalised community recommendations — completely free, funded by partner communities.
Key Features
- Multi-step assessment survey that captures care needs, lifestyle preferences, location, and budget to generate a personalised path within hours
- AI voice intake: Users can complete the entire assessment via a phone call — a Retell AI voice agent conducts the conversation and GPT extracts all required information
- AI-powered matching algorithm that ranks senior living communities based on individual profiles
- Personalised path report including matched communities, financial snapshots, a 90-day action plan, tour guides, and ongoing support resources
- Partner CRM portal for senior living communities to manage leads, track engagement, and communicate with prospective residents
- Admin dashboard for managing surveys, community profiles, blog content (with AI generation via OpenAI), and user overrides
- Email and SMS communication via Resend, Klaviyo, and Twilio — with Handlebars-based email templates
- N8N and Zapier webhook automation for workflow orchestration
- State-by-state community browsing, guides, testimonials, and FAQ
My Role
I built this platform end to end — from architecture to production. The challenge was taking someone's care needs, location, budget, and lifestyle and automatically matching them with the right senior living communities, then delivering a fully personalised recommendation — all without manual intervention. The most interesting part was making the entire journey work through a phone call as well as a web form.
- Integrated Retell AI voice agent with Twilio so users can complete the full assessment via phone — OpenAI GPT extracts 30+ structured fields from the transcript with multi-step validation and retry logic
- Built a geographic matching algorithm using PostGIS and Supabase RPC functions, scoring distance and multi-factor compatibility across 40+ communities simultaneously
- Created an n8n automation pipeline that triggers on intake completion, runs the matching algorithm, and uses OpenAI to generate a personalised HTML/PDF recommendation document — delivered automatically
- Implemented SMS verification (Twilio + JWT), transactional email (Resend), and CRM sync (Klaviyo)
- Architected a Turborepo monorepo with five Next.js apps — survey portal, admin dashboard, partner CRM, user portal, and blog — sharing a tRPC API and Drizzle ORM schema on Supabase