Hateno

Overview

Hateno is not a BI tool. It is a complete, browser-based analytical ecosystem that unites infrastructure, privacy engineering, AI agents, and human-centered design into a single platform for B2B travel intelligence.

Built from the ground up at Salabam Solutions, Hateno consolidates what the market typically keeps separate — IDE, data mining, and AI — into one cohesive environment. It spans four databases, 10+ years of historical data across five international markets (Italy, Spain, France, United Kingdom, United States), and serves both human analysts and autonomous AI agents through a unified privacy-aware API layer.

Multi-Database Architecture

The warehouse is structured around 7 core pillar tables — orders, reservations, services, profiles, flight_segments, events, and ssos — connected through the primary join pattern orders → services → profiles. Together they capture the full lifecycle of B2B travel transactions spanning over a decade.

Privacy-by-Design

Privacy in Hateno is not a post-hoc compliance layer — it is a foundational architectural decision. The system implements a three-tier access model that governs every query, every API response, and every AI interaction:

• Full — Complete data access, restricted to administrators.
• Analyst (default) — Automatic PII transformations applied transparently.
• AI-Agent (strict) — All 119 PII columns completely suppressed. AI never sees personal data.

At the heart of this system sits the Privacy Registry: a YAML-based configuration that maps 24 tables and 119 PII columns across 16 categories of personal information. Six transformation types handle the spectrum of privacy needs:

• Pseudonymize — SHA-256 hashing with deterministic salt, truncated to 8 hex characters.
• Mask — Partial revelation (e.g., email domains preserved).
• Redact — Complete replacement with [REDACTED].
• Generalize — Reduction to less granular levels (birthdates to year, postal codes to first 2 chars).
• Suppress — Complete removal of credentials and tokens.
• Scrub_blob — Recursive cleaning of JSON/XML blobs using 48 JSON patterns and 23 XML tags.

The architecture satisfies GDPR Articles 5 (data minimization), 25 (privacy by design and by default), and 32 (security of processing).

AI-Powered Analytics

Hateno integrates agentic AI workflows where autonomous agents operate on anonymized data through the data-api proxy layer. AI agents receive behavioral instruction files (CLAUDE.md and GEMINI.md) that define analytical scope, available tools, and privacy boundaries.

The platform generates 7 behavioral analytics reports, each designed to surface patterns invisible to traditional SQL queries:

1. Trend di Destinazione — Geographic patterns and "Home Bias" analysis.
2. Insight Comportamentali — Booking habits segmented by traveler type.
3. Tematismi e Correlazioni — Hidden variable relationships.
4. Micro-Nicchie — Unexplored market segments.
5. Geopolitica del Viaggio — International flows and macro trends.
6. Economia del Tempo — Lead time, seasonality, and conversion windows.
7. Paradosso Climatico — Weather correlation with booking behavior.

For ad-hoc exploration, WrenAI provides natural language-to-SQL capabilities, enabling analysts to query the warehouse in Italian without writing SQL.

Data Mining Capabilities

The platform also classifies travelers into 7 behavioral types — solo, couple, family, large_family, group, solo_parent, and extended_family — enabling fine-grained analysis of booking patterns by traveler profile. All visualizations are powered by 20+ interactive ECharts 5.6 charts served through 15+ API endpoints under /api/warehouse/.

Developer Experience

Hateno is designed for zero-friction onboarding. New collaborators go from zero to productive in days, not months. The entire development environment runs in the browser:

• Code-server — Full VS Code IDE accessible through the browser. No local installations required.
• VPN + SSO — WireGuard-secured VPS with Google Workspace single sign-on.
• Vibe-Kanban — Git workflow with isolated worktrees per task, enabling parallel work streams without branch conflicts.
• Hot-reload — Nodemon-powered development server for instant feedback during development.

The onboarding path is structured: Day 1 covers browser login and tutorial introduction. Days 2–3 focus on the Italian analyst guide and ETL transformations. By days 3–5, new analysts explore independently with AI assistance, building their first analyses against real warehouse data.

Documentation as Infrastructure

Documentation in Hateno is not an afterthought — it is infrastructure. The project maintains 218+ Markdown files organized to serve multiple audiences simultaneously: human analysts, developers, and AI agents.

• 167 auto-generated DDL files — Created by dbdump.py, providing schema definitions and sample JSON for every documented table across all 4 databases.
• 19-file Italian analyst guide — Covering glossary, transformations, 8 ETL processes, and all 7 pillar tables.
• 8-chapter hands-on tutorial — Using real warehouse data for practical learning.
• 366-line dashboard manual — With 18 annotated screenshots.

A central INDEX.md hub connects everything, and AI agents use the documentation as their primary context source — making the docs a first-class runtime dependency, not just a reference.

Real-World Insights

The true measure of an analytics platform is whether it surfaces insights that change decisions. Two discoveries from Hateno illustrate this:

The Disneyland Effect

When analyzing family traveler bookings to France, small municipalities — Serris, Chessy, Magny-le-Hongre, Coupvray — consistently dominated the rankings over Paris and other major cities. These are satellite towns of Disneyland Paris. The data revealed that for Italian families, Disneyland is a stronger travel attractor than cultural tourism in major European capitals.

Solo Traveler = Business Traveler

Milan consistently ranked first for solo travelers during working seasons (winter, spring, autumn), yet dropped in summer. This was not leisure tourism — it was business travel: Fashion Week, Design Week, corporate trips. A behavioral pattern made visible only through the 7-traveler-type classification system.