We’re taking the kids to Sydney for the 1-week March school holidays. 8 days covering the city, Blue Mountains, and the Wollongong coast.
Here’s the full itinerary with an interactive map, made by Claude.
How we planned it
We did what everyone does: browsed blogs, asked friends, pieced together a rough itinerary. Then I threw everything at an AI agent (Claude) to refine it.
It reorganized the days, suggested better routing, flagged opening hours, added transport tips.
Then it built a full interactive website with a map so we can pull it up on our phones while we’re there. Day-by-day breakdown, clickable locations, travel times between stops.
That’s where we are with AI now. You give it a rough plan, and it comes back with something genuinely useful.
(Ok, it was guided by human prompting.)
AI and software development
I’ve been using AI heavily at work for coding and development. It’s honestly the most pivotal shift in the 20 years I’ve been building software.
Not a small improvement. But a fundamental change in how I work daily: writing code, debugging, refactoring, navigating unfamiliar codebases.
Now software is cheap and very fast to make.
Case in point: The itineary and interactive map Claude just made probably cost a bomb if you engage a human planner, or use expensive travel tools.
Many will be out of job.
Where AI still falls short
The itinerary planning was great. But there are real gaps.
Finding cheap flights is the big one. Everyone talks about using agents to find the best fares. It still doesn’t work well. They just don’t have reliable APIs to search flights comprehensively.
My wife found our flight using the hidden city technique: you book a flight where your actual destination is a transit stop, then forfeit the final leg. Sometimes this gets you a significantly cheaper fare.
Agents can’t do this. I tried using the Claude Chrome extension to search on trip.com. It couldn’t search fast enough, and it definitely wasn’t thinking creatively about routing. Flight pricing is this constantly shifting mess of airline rules, fare classes, and routing tricks that humans (especially the deal-hunting spouse type) are still better at.
For now.