How I Built Lucia AI: A Custom GPT for Italy Travel Planning (Complete Walkthrough)
Share
I built an AI travel assistant that has helped hundreds of people plan better Italy trips—and it took less than an hour to create. No coding required..
In this walkthrough, I’ll show you exactly how I built Lucia AI using ChatGPT’s Custom GPT builder: the structure, instructions, knowledge base, and testing process. You can use Lucia for your own Italy trip—or apply the same approach to build a Custom GPT for your own niche.
Why I Built Lucia AI
Planning an Italy trip—especially to destinations like the Italian Riviera, Cinque Terre, and Portofino—requires specialized local knowledge. Generic AI tools often provide surface-level advice that overlooks logistics, pacing, seasonal realities, and common traveler mistakes.
I wanted an AI assistant that could offer the same guidance I’d give a friend planning their first Italy trip: realistic expectations, clear trade-offs, and practical logistics. That meant building a Custom GPT with focused instructions and curated travel knowledge.
The result is Lucia AI—a clarity-first Italy travel concierge specializing in the Italian Riviera and designed to help travelers make informed, realistic decisions.
The same clarity-first approach applies beyond AI as well. Many systems appear to work on the surface but fail because the underlying structure isn’t clear — which is exactly what a clarity-first Shopify audit is designed to reveal.
The 4 Essential Fields for Building a Custom GPT
Every Custom GPT is built using four core fields. Here’s how each one works and why it matters.
1. Name
Your GPT’s name should be clear, descriptive, and easy to remember. I chose “Lucia — Italy Travel Concierge” because it’s personal, specific, and immediately communicates purpose.
- Keep it under 50 characters
- Be descriptive, not clever
- Reflect the niche clearly
- Use a subtitle if needed for clarity
2. Description
The description sets expectations at first glance. It should communicate who the GPT is for and how it helps.
“A clarity-first Italy travel concierge, specializing in the Italian Riviera, designed for realistic planning and real-world travel decisions.”
- Front-load the value proposition
- Be specific about the niche
- Avoid generic language
- Keep it under 150 characters
3. Instructions (The Brain)
This is the most important field. Instructions define how your GPT thinks, responds, and prioritizes information. This is where specialization is created.
Key elements to include:
- Tone and voice: calm, practical, and human
- Core principles: clarity over hype, realism over aspiration
- Specialized knowledge: logistics, seasonality, constraints
- Response structure: scannable, actionable outputs
- What to avoid: generic advice, overwhelming itineraries
Lucia’s guiding principles include:
- Prioritize realistic planning over “top 10” lists
- Provide concrete logistics and timing details
- Ask clarifying questions before recommending
- Flag common planning mistakes early
Pro tip: Generic instructions produce generic results. Specific instructions create expertise.
4. Knowledge Base
The knowledge base is what turns a Custom GPT into a true specialist. Uploaded files give your GPT reference material beyond general training data.
Lucia’s knowledge base includes:
- Regional Italy travel guides
- Train logistics and routing notes
- Seasonal travel considerations
- Budget estimates and cost breakdowns
- Local insights and common pitfalls
- Use structured, topic-specific documents
- Include concrete data where possible
- Update files as conditions change
Testing and Refining Your Custom GPT
Your first version won’t be perfect. The goal is to test with real questions and refine based on actual behavior.
- Ask realistic user questions
- Evaluate clarity, usefulness, and pacing
- Adjust instructions to correct issues
- Test edge cases (budget, timing, constraints)
- Repeat until responses align consistently
| Issue | Fix |
|---|---|
| Too generic | Add more specific instructions and examples |
| Too long | Require concise, scannable outputs |
| Missing details | Upload better knowledge files |
| Too formal | Adjust tone guidance |
How Travelers Use Lucia AI
Lucia is designed to work like a conversation with a knowledgeable travel advisor.
- Travelers start with a broad question
- Lucia asks clarifying follow-ups
- Recommendations account for pace, budget, and logistics
- Plans are refined iteratively through conversation
ChatGPT vs. Lucia AI (Real Comparison)
Using the same prompt, the difference is clear.
ChatGPT: broad destination lists, minimal logistics, unrealistic pacing.
Lucia AI: clarifying questions, realistic base locations, timing guidance, and seasonal awareness.
Build Your Own Custom GPT (Checklist)
- Define a focused niche
- Write clear, specific instructions
- Upload relevant knowledge files
- Test with real questions
- Refine continuously
Total setup time: 1–2 hours for a functional GPT.
Common Questions
No coding required. Custom GPTs are built entirely through text fields and uploads.
Monetization options: lead magnets, services, products, or OpenAI revenue sharing.
Updates: Instructions and knowledge can be edited anytime.
Key Takeaways
- Custom GPTs are easy to build
- Instructions are the most important component
- Specificity creates value
- Testing and iteration matter more than perfection
Next Steps
Planning an Italy trip?
Try Lucia AI for free →
Want help building your own GPT?
Explore AI Services →
Need a starting point for Italy planning?
Download the Italy Trip Planning Guide →
Questions? Drop a comment below or book a consultation.