Today, I was blown away using ChatGPT to suggest a two-week Portugal/Spain trip plan with my wife for next February.
I started prompting with…
Please ignore prior instructions.
As an expert Portugal and Spain travel agent, please recommend a romantic 12-day trip during February.
The travelers are my wife and I, visiting for the first time.
We prefer to walk or ride the train.
Please start from the Porto or Seville airport and return to the arrival airport.
Keep to a travel-based response.
Please include a download link for the output.
Max Tokens: 150
Temperature: 0.6
Top-p: 0.9
Frequency Penalty: -0.3
Presence Penalty: 0.2
Tone: Warm and Affectionate
Writing Style: Conversational with a dash of Formality

The ChatGPT settings at the bottom are based on asking…
Please ignore prior instructions.
{type} = "romantic travel agent"
As a senior ChatGPT prompt engineer, please suggest ChatGPT {options} relevant for the persona type: {type}. Output the suggestion in plain text in a ready-to-use format. Afterward, provide a brief description.
{options} = "Max Tokens, Temperature, Top-p, Frequency Penalty, Presence Penalty, Tone, Writing Style"

Learn more about ChatGPT settings.
While working towards that first prompt, I ended up with several itineraries and wanted to combine them.
As a romantic travel agent, please combine the suggested itineraries below the ***** marker into a 12-day trip. Ensure that the trip starts and ends in Porto.
*****
COPY & PASTE OF ITINERARIES
Once compiled, I wanted to add a unique daily experience.
Thank you for the singular itinerary.
For once each day, suggest a unique new experience beyond going someplace. For example, a food or drink to try or a place to do or learn something.

However, a new itinerary was given, so I merged the new with the old.
Please combine the previous itinerary and suggested experiences together.

While feeling pretty good about the planning and wanting more details to help my wife make decisions, I asked.
For each city and activity, please add a description.
However, I lost daily details. Also, I began feeling cumbersome with copy and paste of the itinerary as my refinements drew on.
So, I saved the primary itinerary as a plain text file and used ChatGPT 4’s code interpreter to upload the information when asking for more details.
As a romantic travel agent, for the uploaded itinerary, add to each day a new experience, like something to eat, drink, do, or learn.
For each city, add a description.
For each place, add a description.
For each activity, add a description.

Wowza, impressive ideas, and why do them.
Let’s download them.
Please download above as a nicely formatted PDF.
Later in the day, I wondered about adding references to where to learn more or buy tickets.
As an expert Portugal and Spain travel agent, for the uploaded itinerary, please references on where to learn more or purchase tickets for the activities.

In hindsight, I botched that prompt.
The response ended up being more of a city-based reference. Also, while the suggested links worked, they weren’t specific enough for me.
For example, the Porto Wine Cellars link is for visitportoandnorth.travel than porto.travel/port-wine-cellars.
Therefore, I manually merged the information, found informative links, and shifted to inline referencing rather than a separate section.
However, here’s a possible prompt rewrite for adding URLs within the itinerary.
As an expert Portugal and Spain travel agent, for each activity of the uploaded itinerary, please add inline URLs on where to learn more or purchase tickets.
Wrapping Up

Within a few hours of effort, I created an adventure for Huan and me while learning much about Portugal and Spain. And the suggestions stayed relevant, such as walking around, not driving.

ChatGPT has prevented days of headaches and stress in creating a travel plan to kick off deeper discussions and feel excitement.
When I give myself the freedom to be curious and dig deeper, ChatGPT is an incredible educational tool and productivity enhancer.
Where might ChatGPT help you explore next?

Leave a reply to Michael Cannon Cancel reply