First Assignment: Vibe Code Your Own Vacation.
🏝️ (Because Why Not?)
Hey Agents-in-Action crew 👋
Okay, okay. I know what you're thinking.
We just spent the last four days getting deep in the weeds about tool calling, memory systems, ReAct patterns, and — thanks to Day 5 with Areej — how to make agents that can actually see things and not just hallucinate about them.
But you know what? Let's change gears.
Not because we're slowing down (we never slow down), but because I want to show you something that'll make everything we've covered suddenly click into place.
Welcome to your first unofficial assignment:
👉 Vibe code your own vacation.
Yeah, you read that right. We're building a travel agent. With AI. That actually works.
✨ What is "vibe coding," anyway? (And why does it sound made-up?)
Look, you've heard of prompt engineering. You've seen people duct tape ChatGPT to everything and call it "AI automation." You've probably rolled your eyes at least seventeen times this week at LinkedIn posts about "revolutionary AI workflows."
But vibe coding is different — and it's the secret sauce we use in my full course to move fast, build flexibly, and let the agent do the heavy lifting while you focus on the fun stuff.
Instead of spending your weekend writing 47 different if-else statements to handle every possible way someone might ask for a "cozy cabin with good WiFi" — you design flows that just... get it. They vibe with the user. They're smart enough to figure out what you probably meant, but explicit enough that you can debug them when they inevitably do something weird.
(And today, we're going to do that with the most important use case of all: ✨finding your dream vacation rental✨.)
🧠 The Assignment: Build a "Vacation Agent" That Actually Understands You
Here's what we're building: a semantic vacation search agent that doesn't just filter by price and location (boring), but actually understands the vibe you're going for.
The Stack:
🏡 Airbnb's MCP Server (the fancy new Multi-Control Protocol stuff)
🔎 Natural language → structured search (this is where the magic happens)
⚙️ n8n webhooks (because sometimes the best tool is the one that just works)
🎨 A frontend that doesn't suck (built with Lovable.dev, naturally)
The Experience: You tell it something like:
"I want a place in Bali where I can work remotely but also feel like I'm actually on vacation. Something with a pool, good WiFi, and maybe some Instagram-worthy views. Budget is flexible but don't go crazy."
And instead of throwing back 847 random listings, it:
Understands what "work remotely but feel on vacation" means
Translates "Instagram-worthy views" into actual search parameters
Finds properties that match your actual vibe, not just your filters
Explains why each recommendation makes sense for you
This isn't just a fancy UI wrapper around Airbnb's existing filters. This is actual agent behavior — interpretation, reasoning, and personalized recommendations.
🧰 What You'll Actually Learn (Spoiler: It's Not Just About Vacations)
Even though this sounds like a fun weekend project, you're actually touching every core concept we build on in the full course:
🧠 Natural Language → Structured Intent
You'll learn how to take messy human requests and turn them into clean, actionable parameters. This is the foundation of every business automation agent you'll ever build.
🔌 Real-World API Integration
Forget toy examples. You're connecting LLMs to live data sources, handling rate limits, managing API keys, and dealing with all the fun stuff that happens when rubber meets road.
🎭 Multi-Agent Orchestration
One agent handles the search logic. Another handles the recommendation ranking. A third manages the user interaction. Welcome to the real world of AI systems.
🎨 Vibe-First Interface Design
Not every AI tool needs to look like a chatbot. You'll learn how to design experiences that feel intuitive and handle ambiguity gracefully (hint: this is how you make AI tools people actually want to use).
🤫 The Real Secret (Don't Tell Anyone)
This project? It's basically a Trojan horse.
On the surface, you're building a travel tool. But under the hood, you're learning the same patterns that power:
Research assistants that understand your domain and find relevant papers
Sales agents that qualify leads based on fuzzy criteria
Support bots that actually solve problems instead of just apologizing
Content generators that match your brand voice and audience preferences
The only difference is the domain. The agent architecture? Identical.
🚀 Ready to Actually Try It?
Before we dive into building, I want you to experience what we're aiming for.
I've built a working version of this exact agent:
👉 Try the Chrono Voyage AI Travel Agent
Go ahead, give it a real travel query. Something you'd actually want to book. See how it interprets your request, finds options, and explains its reasoning.
Notice:
How it handles ambiguous requests ("somewhere relaxing")
The way it prioritizes recommendations based on your specific needs
How the explanations help you trust (or question) its suggestions
This is what vibe coding looks like in practice. It's not perfect, but it's useful — and that's often more important.
Here’s the website in action:
Here’s the n8n workflow:
🎓 Want to Build Systems Like This? (Of Course You Do)
If you're playing with the travel agent and thinking "okay, this is actually pretty cool, how do I build stuff like this for real problems..."
You're ready for the full experience.
Building Gen AI Agents: Beyond the Hype starts July 2nd at 7PM PT.
We go deep on everything:
ReAct and ReWOO patterns (the stuff that actually works in production)
Multi-agent pipelines that don't fall over when things get complex
Google's Agent Development Kit (ADK) and MCP workflows
How to add guardrails without killing the magic
Real projects: your own Perplexity clone, semantic caches, private reasoning agents
Plus the stuff nobody talks about:
When not to use agents (seriously, this will save you months)
How to debug agent behavior (spoiler: it's an art form)
Making agents that surprise you in good ways
Flirting with your own startup? Several students have turned course projects into fundable ideas
Use this link for 20% off (expires in 3 days): 👉 Join the July cohort
🏖️ Your Mission (Should You Choose to Accept It)
Whether you end up in the full course or not, I challenge you to build something this week.
It doesn't have to be a travel agent. Maybe it's:
A restaurant finder that gets your dietary restrictions and your mood
A gift suggester that knows your budget and your weird relatives
A workout planner that adapts to your actual energy levels
A book recommender that understands "like Harry Potter but for adults who pay taxes"
The pattern is the same: natural language input → intelligent interpretation → personalized results → clear explanations.
Start small. Start weird. Start with vibes.
And hey, if you do build something cool, hit reply and show me. I love seeing what people create when they stop overthinking and just... build.
So whether you're planning your next vacation, automating your side hustle, or just trying to make AI do something actually useful...
The vibes are in motion.
See you in the mountains (or the course, or both) 🏔️✨
—Hamza LinkedIn | Substack | Email me weird ideas
P.S. — Yes, I actually used the travel agent to plan my last three trips. No, I'm not sponsored by Airbnb. Yes, it found some genuinely amazing places I never would have discovered otherwise. The algorithm knows me better than I know myself at this point, and I'm only slightly concerned about that.
Don't forget to try the Chrono Voyage AI Travel Agent — seriously, it's more fun than it has any right to be.
You're receiving this email because you're part of mailing list , and you've attended, registered for, or were invited (by a co-host, perhaps) to our events on MAVEN or other forms. These emails are the only way to reliably receive updates from us. We don't spam or sell your information. If you prefer not to receive our messages, simply "unsubscribe" below and we'll respect your desires. We endeavor to keep our list clean and only message people who want to hear from us. (Please do not mark messages as spam, as this makes it harder for those who want updates to receive them. Thanks!)
 | A guest post by
|