Destination pages, package pages, Markdown files, JSON files, and metadata were spread across the site.
AI TRAVEL ADVISOR
Turning a content-rich travel website into an intelligent conversational advisor.
A travel business had valuable destination pages, packages, itineraries, offers, FAQs, and editorial content. Boldcraft turned that published website content into an AI-powered advisor grounded in the company's own data.
THE PROBLEM
Strong content existed, but customers still had to hunt through it.
Visitors asked about budget, dates, mood, inclusions, destination fit, and travel style in one sentence.
The AI system needed a repeatable way to stay aligned whenever the website was published.
THE STRATEGY
The website manifest became the source of truth.
Structured data
Exact travel facts
Package names, prices, dates, destinations, durations, inclusions, tags, categories, availability, and metadata.
SQL database
PriceDatesDurationAvailabilityInclusionsTags
Unstructured data
Travel meaning
Destination descriptions, itinerary narratives, travel notes, FAQs, highlights, and long-form Markdown content.
Vector database
DescriptionsItinerariesFAQsHighlightsTravel styleMood
INGESTION ARCHITECTURE
01
Website published
The travel site publishes destination pages, package data, itineraries, offers, FAQs, and editorial content.
02
Manifest created
A manifest lists the JSON and Markdown files that represent the latest published website content.
03
Content normalized
The ingestion layer reads files, cleans fields, and separates exact facts from long-form narrative content.
04
Two stores updated
Structured package facts go to SQL. Descriptive Markdown chunks become embeddings in a vector database.
05
Advisor answers
The Mastra agent combines both sources to recommend trips, compare options, and answer follow-ups.
HOW THE ADVISOR ANSWERS
Query type Example Retrieval Response behavior
Structured lookupShow me trips under Rs. 80,000 for 5 days.SQL databaseFilters exact records by budget, duration, and availability.
Semantic discoveryWhere should I go for a calm beach holiday?Vector databaseFinds destination and itinerary content that matches the intent.
Hybrid recommendationSuggest a premium 6-day trip with beaches and good food.SQL plus vectorCombines package filters with descriptive travel context.
ClarificationPlan something nice for December.Agent-led follow-upAsks for missing details like budget, duration, destination, or style.
ComparisonWhich option is better for a relaxed family trip?Ranked hybrid resultsCompares options using facts and experience descriptions.
BEFORE AND AFTER
Before After
Customers browsed multiple pages manually. Customers asked questions through chat.
Search depended heavily on keywords. The advisor understood intent and context.
Package data and editorial content were disconnected. Structured and unstructured knowledge worked together.
Website updates were not automatically AI-ready. Published content could be ingested into the AI system.
Discovery was passive. Discovery became conversational and guided.
TECHNICAL ARCHITECTURE
The AI Hub layer behind the travel advisor.
This is how the published website, ingestion pipeline, SQL records, vector search, Mastra tools, ranking logic, and chat experience work together.
ARCHITECTURE LAYERS
LayerPurpose
Website content sourceUses the published website manifest as the source of truth.
Ingestion pipelineReads JSON and Markdown files, cleans them, and prepares them for AI usage.
Data storage layerStores structured facts in SQL and unstructured content in a vector database.
Tool layerExposes dedicated tools for structured lookup, semantic search, comparison, ranking, and response preparation.
Agent orchestration layerUses Mastra to understand the query, call tools, compare results, rank evidence, and generate answers.
Chat experiencePresents the AI advisor as a conversational interface on the website.
CONTENT PROCESSING
Content typeExampleStorageUsed for
Structured dataPackages, dates, prices, destinations, metadata, inclusionsSQL databaseExact filtering, factual answers, comparisons
Unstructured dataMarkdown pages, descriptions, FAQs, itinerary notes, travel narrativesVector databaseIntent matching, semantic discovery, contextual reasoning
AGENT TOOL FLOW
01
Input
Customer query
A traveller asks a natural-language question.
02
Agent
Mastra agent
The advisor receives the message and prepares tool use.
03
Reasoning
Intent + constraints
Budget, date, mood, destination, or comparison intent is identified.
04
Retrieval
SQL + vector tools
Structured lookup and semantic search run as needed.
05
Evidence
Compare evidence
Facts and content matches are checked together.
06
Ranking
Rank best matches
The strongest options are ordered by fit and confidence.
07
Output
Grounded response
The chat answer cites real package and content evidence.
RANKING LOGIC
SignalSourceWhy it matters
Exact matchSQL databaseConfirms facts like price, duration, destination, category, or availability.
Semantic matchVector databaseCaptures intent, mood, travel style, and descriptive relevance.
Constraint fitSQL databaseChecks whether the option satisfies user constraints.
Context fitVector databaseChecks whether the experience matches the user's stated preference.
ConfidenceCombinedHelps the agent choose whether to answer directly or ask a follow-up question.
EXAMPLE QUERY
"Suggest a premium 5-day beach holiday with good food."
- Call structured lookup for 5-day beach-related packages.
- Call semantic search for premium stays, coastal experiences, food, and relaxed travel.
- Compare both result sets.
- Rank trips that satisfy duration and match travel style.
- Generate a concise recommendation with supporting reasons.
Boldcraft helped a travel business transform its published website content into an AI-powered travel advisor. The ingestion layer converted JSON and Markdown files from the website manifest into structured SQL data and semantic vector data. Mastra powered the advisor layer on top, orchestrating retrieval, ranking, and grounded responses so customers could ask questions, compare options, and discover relevant trips through conversation.
TALK THROUGH A TRAVEL AI SYSTEM →