This is a submission for the AI Agents Challenge powered by n8n and Bright Data
What I Built
I built Ceylon Journey Partner – Real-Time AI Travel for Sri Lanka, an AI-powered travel planning workflow that creates a personalized itinerary for tourists visiting Sri Lanka.
The workflow solves the common problem of overwhelming travel research by automating the process of collecting hotel options from multiple platforms (Booking.com, Airbnb, Agoda) and combining them with user preferences such as travel dates, budget per day, destinations, and activities.
It ensures travelers get a balanced plan that maximizes good weather, minimizes travel time, fits within budget, and matches personal interests.
Preview Workflow
Demo
n8n Workflow
GitHub Gist - Ceylon Journey Partner
Technical Implementation
The workflow is implemented in n8n with the following components:
System Instructions (AI Agent Prompt):
You are a Sri Lanka travel planning expert.
Using the provided hotel data (Booking.com, Airbnb, Agoda), travel country, destination cities, travel dates, budget per day, traveler count (adults, children, infants, pets), and interested activities:
Return a complete travel plan that:
- Covers ALL provided cities between the given dates.
- Suggests daily activities near each city based on the interests provided.
- Recommends at least one accommodation option in each city within the budget.
- Estimates daily costs (accommodation, activities, transport).
- Highlights potential risks (weather, closures, peak crowd times).
- Suggests 1–2 backup options if risks exist.
Model Choice: Google Gemini 2.5 Flash
(via n8n AI Agent node)
Memory: Stateless (each itinerary generated fresh from real-time inputs)
Tools Used:
-
Webhook Trigger
- Entry point for user input.
- Captures country, destination cities, travel dates, budget per day, traveler count (adults, children, infants, pets), and interests.
-
Collect Inputs
- Normalizes the JSON payload for downstream processing.
-
Hotel Data Scraping with Bright Data
- Three Bright Data Verified Nodes + custom Code nodes filter hotel listings from:
- Booking.com dataset (
gd_m5mbdl081229ln6t4a
) - Airbnb dataset (
gd_ld7ll037kqy322v05
) - Agoda dataset (
gd_m837ssst155rq3a1xo
)
- Booking.com dataset (
- Each platform uses filter queries based on country and destination cities.
- Snapshots are created, checked for readiness, and retrieved.
- Three Bright Data Verified Nodes + custom Code nodes filter hotel listings from:
-
Snapshot Management
- Snapshot Status node checks if Bright Data snapshot is built.
- If + Wait nodes ensure retry until dataset is ready.
- Switch Node maps dataset IDs to Booking, Airbnb, or Agoda.
-
Hotel Data Aggregation
- Aggregate nodes collect and unify hotel listings per platform.
- Merge node combines Booking, Airbnb, and Agoda data into a single dataset.
-
User & Hotel Data Merge
- User preferences and hotel datasets are aggregated into a structured input for the AI agent.
-
AI Agent Node (Google Gemini 2.5 Flash)
- Prompted as a Sri Lanka travel planning expert.
- Generates a day-by-day itinerary that:
- Covers all destination cities within travel dates.
- Suggests activities aligned with user interests.
- Recommends budget-matching accommodations (Booking.com, Airbnb, Agoda).
- Estimates daily costs (hotel, transport, activities).
- Highlights risks (weather, closures, crowds).
- Suggests backup options.
- Output format: Gmail-compatible HTML (tables, headings, booking links).
-
Email Delivery
- Final travel plan sent to the user via the Gmail Node.
- HTML itinerary is styled and directly viewable in email.
Bright Data Verified Node
Bright Data’s Verified Node was used to scrape hotel availability
, pricing
, and location details
from Booking.com, Airbnb, and Agoda in real-time.
This ensures itinerary recommendations are always up-to-date with accurate prices and availability.
Journey
I wanted to solve a real-world tourism problem in Sri Lanka — how to plan trips without juggling dozens of tabs.
- The first challenge was merging heterogeneous data. I overcame this by using n8n’s Merge Node and custom function nodes to structure everything.
- Next, I crafted an AI Agent prompt that balanced flexibility with structure, so itineraries are clear but still personalized.
- Finally, integrating Bright Data ensured the agent works with real-world, ever-changing data rather than static APIs.
Key challenges I overcame:
- Merging different hotel data sources into a single dataset
- Normalizing budget and traveler preferences into a scalable filter
- Generating itineraries that balance weather, budget, and travel time
- Automating the entire process into a single n8n workflow
Through this, I learned how powerful Bright Data + n8n can be when combined with AI — enabling real-time decision-making for travelers.
🚀 With Ceylon Journey Partner, anyone can instantly generate a real-time, AI-powered Sri Lanka itinerary — combining the magic of travel with the reliability of structured data.
Top comments (0)