On call · family

Travel Concierge

An AI concierge that plans our trips, builds the itineraries, and asks us how it went.

Why it exists

We belong to a vacation club with dozens of properties, which sounds relaxing and is actually a part-time job: booking windows open and close, availability vanishes, and someone has to track it all, build the day-by-day plan, and remember what worked last time. I gave the job to software. Now the family discusses trips with a concierge that already knows our history and our tastes.

What it does

Watches for openings. Monitors availability for the places we want, and flags windows that fit the family calendar.
Plans conversationally. We tell it what kind of trip we're imagining; it comes back with options, trade-offs, and a recommendation.
Builds real, living itineraries. Day-by-day plans with flights, weather, and costs — synced in real time and reorderable by everyone on the trip at once.
Runs the debrief. After each trip it asks what we loved and what we'd skip, and the answers shape the next plan. It gets better every trip.

Availability monitoring across the club's properties — demo data

Multi-year booking strategy, generated by the concierge — demo data

A live shared itinerary — synced and reordered in real time — A live shared itinerary — synced in real time, edited by the whole family at once

Status	In service for family trips
Cadence	On call; availability monitoring runs continuously
Built with	Node.js + PostgreSQL with an AI concierge layer, built by my agent fleet

The honest catch

The concierge is only as good as its memory of what the family actually enjoyed — and collecting honest post-trip feedback from busy humans turns out to be the hardest engineering problem in the system.