AI3 — Owner Insight Assistant Specification
Surface: Operator Desktop (Reports module) + Control Plane (chain operators) Phase: P2 / R3 Feature flag:
owner-insight-assistantCompetitive reference: Duetto AI, Lighthouse, OTA Insight, Notion AI analytics
1. Overview
The owner insight assistant is a natural-language analytics tool for property owners and general managers. Instead of navigating complex report dashboards, owners ask questions in plain language and receive instant answers with charts, tables, and AI-generated insights.
Example interactions:
- "What was our RevPAR last month compared to the same period last year?"
- "Which room type had the highest upsell conversion in Q1?"
- "Why did occupancy drop on 15 March?"
- "Generate a revenue summary for my board presentation."
2. Entry point
Operator Desktop Reports module: A "Ask a question" input bar at the top of the Reports section.
Control Plane (chain operators): Persistent "Insight" tab across the multi-property dashboard.
3. Query types
| Query type | Example | Output |
|---|---|---|
| Metric lookup | "What's my ADR this week?" | KPI card with value + trend |
| Comparison | "Compare RevPAR: this month vs last year" | Line chart |
| Anomaly investigation | "Why did occupancy drop on 15 March?" | Prose explanation + contributing factors list |
| Ranking | "Which room type generated most revenue in Q1?" | Bar chart + ranked table |
| Forecast | "What's the projected occupancy for next weekend?" | Forecast chart (Phase 3 — requires ML model) |
| Report generation | "Summarise last quarter for my board" | Exportable markdown/PDF report |
| Cross-property (chain) | "Which property had the best NPS last month?" | Ranked table across properties |
4. Response anatomy
┌─── Owner Insight ─────────────────────────────────┐
│ Q: "What was our RevPAR last month vs same period │
│ last year?" │
│ │
│ 📊 [Line chart: Apr 2026 vs Apr 2025] │
│ Apr 2026: $87.40 RevPAR (+12% YoY) │
│ Apr 2025: $78.00 RevPAR │
│ │
│ AI commentary: │
│ "RevPAR improved 12% year-over-year, driven by │
│ a 4% increase in ADR ($104 → $108) and a 3.5pp │
│ occupancy improvement (78% → 81.5%). The bump │
│ coincides with Nawroz holiday (20–22 Apr) which │
│ was not a factor last year." │
│ │
│ [Export as PDF] [Add to presentation] [Drill in] │
│ │
│ Sources: Reservation data · Folio revenue · │
│ Holiday calendar (Nawroz 20 Apr) │
└────────────────────────────────────────────────────┘
5. Chart suggestions
If the owner's query is ambiguous about visualisation, the AI suggests the most appropriate chart type:
Q: "Show me booking trends by source"
AI: "Here's a breakdown by booking source (OTA vs direct vs walk-in):
[Bar chart] — or would you prefer a [line chart] to see trends over time?"
[Show bar chart] [Show line chart]
The owner can select or describe what they want in natural language.
6. Export to deck
"Add to presentation" generates a slide-ready PNG (16:9, white background, property branding, Melmastoon watermark) of the chart + AI commentary. Exported via export-service to the operator's email.
Phase 3: PowerPoint/Google Slides native export.
7. Data access and security
The insight assistant has access to only the requesting operator's property data. Chain operators see their connected properties only.
Data accessed:
- Reservations (anonymised guest data for analytics; no PII)
- Folio revenue (aggregated)
- Occupancy and ADR metrics (from
analytics-service) - Upsell conversion rates
- Review scores
Not accessible:
- Individual guest PII (names, passport numbers)
- Other tenants' data
- Platform-level billing data
8. Explainability
Every AI-generated insight includes a "How did we calculate this?" link:
- Opens a methodology card explaining the metric definition (e.g., "RevPAR = ADR × Occupancy rate")
- Lists the date range and filters applied
- Shows the raw numbers used in the calculation
This prevents "AI magic" trust issues with owners.