This is Part 1 of the Beyond ChatGPT series — about what AI can actually do, not just think. Most people use AI to get answers. This series is about using it to get things done.
I'm just home from an amazing 12-day cruise — Amsterdam, Scotland, and Iceland. Midnight sun, geysers, waterfalls, the kind of landscapes most people dream of.
My travel partner Cathy and I started planning the way most people do: fifteen browser tabs, a Google Doc that became a graveyard of good intentions, two people with different priorities texting at all hours. Overwhelming — until AI took over the heavy lifting.
This post is about exactly what that looked like. Not AI answering questions. AI doing work.
AI That Thinks vs. AI That Does
Most of us use AI like Google — ask a question, get an answer, close the tab. Useful. But only a fraction of what's available.
There's a version of AI that works inside your actual files — editing your spreadsheet, building maps in your browser, pushing a navigable list of every landmark to your phone. It takes the raw plan you described and turns it into something you can actually use on the ground.
The gap between "I asked AI about this" and "AI built this thing I'm now using" is smaller than most people think. Less conversation. More execution.
ChatGPT gave me information. Codex — ChatGPT's tool for working inside files — built me a logistics system. Here's how.
Step 1: ChatGPT Did the Research
I gave ChatGPT our cruise itinerary for Iceland — each port, when we'd dock and depart, what we cared about. Two women in our sixties, fit but on vacation. Waterfalls, lava fields, Instagram-worthy photo spots — not expensive guided tours with matching lanyards.
ChatGPT gave me an editorial briefing. Not everything we could do — a curated shortlist of why this place over that one, based on our interests and time.
I also uploaded documents directly: the cruise line's PDF excursion guide, my planning notes in Word (flights, stuff already booked, etc.) No retyping or tab-switching — I handed the files to ChatGPT and asked it to work on them: summarise the options, flag conflicts with our priorities, tell me what was worth reading. This saved me hours.
ChatGPT gave me research. It didn't give me a plan. For that, I needed something else.
Step 2: Codex Turned My Ideas Into a Working Tool
I had a Word doc and table — ports, dates, rough notes — the kind you build optimistically and then ignore. I handed it to Codex.
Codex is ChatGPT's tool for working directly inside your files. Not summarizing them. Not copying from them. Actually editing them — on your computer, in real time.
I asked it to pull in my ChatGPT research and organize everything by port, by priority, by time needed. Add driving distances. Flag when we had to be back on the ship. It built me a document we could actually run a trip from — with a rental car across Iceland for three days, that mattered.
Then I asked it directly: Is this too much? Are we driving more than experiencing? The answer came back clearly. We adjusted. That conversation happened inside the document — no copy/paste, no switching apps.
Insider tip — the two-pass. I now do this with many things AI produces. Give the output to a second conversation and ask: What's wrong with this? What did it miss? Is this realistic? One minute. Moves you from a generic AI answer to something you actually trust.
Step 3: ChatGPT Found the Ship From the Port Registry
Cruise line information is, to be generous, approximate. You get a port. What you don't get is which dock, whether you're tendering (taking a small boat to shore instead of walking off), where the tender drops you, and how much time that actually costs you.
For one of our ports, the information I had was wrong. For others, just incomplete.
So I asked ChatGPT to look it up. It cross-referenced our ship against maritime port registries — the databases that track vessel movements and berth assignments. It came back with the specific dock, confirmed tendering, and told me exactly where we'd be dropped off. From there I could plan realistically: distance to town, transport options, actual time lost to transit.
This is AI pulling from operational data — not a travel blog. The kind of thing a seasoned agent used to know.
Step 4: Codex Built My Google Maps List and Routes
This is the step that changed how we navigated on the ground — and the one people ask me about most and my best personal learning from this AI work.
Once the itinerary was solid, I asked Codex to do two things in my browser:
First: build a Google Maps List. Every named place — landmark, shortlisted restaurant, car rental location, meeting point — added to a single saved List in Google Maps. A List is Google Maps' native way to save and organize locations. It synced instantly to my phone. On any given day, I opened the List, tapped a pin, and had directions, opening hours, and reviews right there. No searching. No "what was the name of that place?" standing in the rain in Reykjavik.
Second: build day-by-day routes. Codex created route maps for each day showing the sequence of stops in order. I could see at a glance whether a morning's plan made geographic sense, how long the driving would actually take, and whether we'd designed a day that doubled back on itself. Routes helped me validate timing and geography before we left the ship.
The List was what I used every day on the ground. The routes were what made the planning real.
Step 5: Two Research Styles, One Trip
Cathy and I both used ChatGPT — but differently, because we think differently.
Cathy asked it to generate images of places she was considering. Visual, instinctive — she wanted to see it before committing. I built timing logs and route structures. Analytical — I needed it to work on paper first.
Then we shared our chat transcripts with each other's AI: Given both sets of priorities — what's an itinerary that works for both of us?
AI isn't trying to win. It looks at two wish lists and builds something that honours both. We landed on itineraries neither of us would have designed alone. We had a solid plan and the data to make decisions — and we knew the plan could flex day by day if needed.
What This Actually Is
Three tools. One process.
- ChatGPT — research, briefings, document analysis
- Codex — spreadsheet editing, map building, route creation, all in the browser
- Google Maps — the List and routes on the ground
Not three separate tools. One workflow: understand, structure, deploy.
AI isn't just for answering questions. It's for doing work — inside your files, inside your browser, on your phone.
A Few More Things Worth Trying
AI is already inside Word and Google Docs. Copilot in Word, Gemini in Google Docs — highlight text, ask AI to rewrite, summarise, or expand it. Right there in the document. No copy/paste.
Packing lists by destination. Ask ChatGPT to generate one for your specific locations, dates, and activities. More useful than generic because it knows the context.
Real-time translation. ChatGPT's camera reads menus. Quietly useful in Iceland.
Replanning on the go. Plans change — weather, closures, you're running late. Paste your remaining day into ChatGPT and ask it to replan from where you are.
Try This Before Your Next Trip
Start with one step.
Research: Tell ChatGPT where you're going, who you're with, and what kind of traveller you are. Upload your cruise or tour PDFs and ask it to pull out the decisions you actually need to make. Read this first: How to Talk to AI: The Four-Part Prompt Formula.
Planning: Hand your spreadsheet to Codex. Ask it to add structure, driving times, and flags. Ask whether the plan is realistic. Under twenty minutes.
Navigation: Ask Codex to build a Google Maps List from your itinerary and day routes for each port. The List is what you'll live in on the ground. The routes are what make the timing and geography real before you arrive.
That last step is the one I get the most questions about — so I've written it up in full, with the exact prompts I used. DM me the word TRAVEL and I'll send it to you.
Consider it decoded.
Written by Amanda · Midlife TechCurious
Next in the Beyond ChatGPT series: AI and your creative life — same idea, different territory. Not AI thinking for you. AI doing something alongside you.