Networking CRM + Family Tree

My whole family — 212 people, 359 relationships — mapped, related, and driven by AI.

May 24–31, 2026
  • Next.js
  • Prisma
  • Leaflet
  • OpenAPI
  • GPT Actions
Interactive family tree centered on Dan Maccaba, other names blurred
The tree view — 212 people, connected. Names blurred for their privacy; I'm the green card.
212family members
359relationships
38places geocoded on the map

Two apps that became one project: a personal CRM for the people in my professional orbit, and a family tree for the 212 people in my personal one.

The family side is the heart of it. Every relative, every relationship — 212 people, 359 relationships (those numbers come straight from the database) — with a tree view for how everyone connects and a map showing where the whole family lives, 38 geocoded places across Israel and the world. Grandparents love it, which is the only user metric that matters here.

The map — every contact in the network placed across the world

The two datasets nest inside each other: everyone on the family tree is also in the networking CRM — but not everyone in the CRM is on the family tree. The CRM is the superset, holding colleagues, friends, and professional contacts alongside the family; the tree is the family-only view with its own relationship graph.

The networking CRM — 212 people, names blurred

The CRM side carries the project’s most interesting engineering choice: I wrote a 744-line OpenAPI specification by hand, 13 authenticated endpoints, and pasted it into a ChatGPT Custom GPT. The result: I talk to my CRM. “I met Noa at the conference, she works on medical imaging, remind me to write to her in two weeks” — and it’s all filed.

This was May 2026, early in my streak of wiring AI assistants into everything I build. It’s where that pattern clicked.

Want someone who builds like this?

I'm open to opportunities. Tell me what you're building — I answer every email.