AI Concierge
Web & WhatsApp chatbot · Developed by Octave · 4 versioned releases
Vision: AI-powered guest-facing chatbot on website and WhatsApp. Progresses from informational MVP (V0.1/V0.2) to full booking, payment and personalisation (V1.0/V1.5).
M1
MVP / GTM Launch
V0.1 (Web) + V0.2 (WhatsApp) go live
26 May 2026
M2
Full Upgrade Launch
V1.0 (Web Booking) + V1.5 (WhatsApp Advanced) go live
11 Sep 2026
V0.1 — Web MVP
In Progress
- CMS-based knowledge retrieval for informational queries
- Menu display via Canvas API (PDFs & images) Done
- Dynamic URL linkout for room booking Done
- Google Maps distance to nearby places WIP
- EatApp iFrame for restaurant booking Done
- Spa booking linkout Pending
- Call centre contact info retrieval Done
- Basic usage analytics Pending
- Cloud provisioning & production migration WIP
- Internal testing → UAT → Sign-off → Website integration via Tag Manager
Team: Pamith, Mansitha, Hirusha, Ransika/Naduni · PM: Sayuni
V0.2 — WhatsApp MVP
Pending
- Replicate V0.1 features #1–7 as text-only via 360SMS
- WhatsApp usage analytics (360SMS/Meta + Octave session analytics)
- Configuration → Internal Testing → UAT → Sign-off
- Connect 360SMS app with WhatsApp business number
Lead: Mansitha · Dep: 360SMS endpoint mapping (target 24-Mar) · ~3 weeks once received
Note: CHML will not launch with SF Journeys at MVP — potential 3–4 week downtime risk.
V1.0 — Web Booking
Not Started
- Sabre SynXis direct reservation creation & modification (~12 weeks dev)
- RedDot payment gateway via SynXis linkout
- GHA membership validation (recognition & payment)
- Sabre certification + mandatory monitoring/evaluation
- Complete UAT + third-party VAPT security testing (~4 weeks)
Lead: Mansitha & Hirusha · Clarity: Unclear on several items — depends on Sabre sandbox & partner meetings
V1.5 — WhatsApp Advanced + Analytics
Pending
- 360SMS full integration development (~6 weeks by 360SMS)
- Frontend design for WhatsApp components
- Salesforce personalisation — e.g. "Hi <Sadeep>" via SF data retrieval
- Platform analytics dashboard (Web & WhatsApp)
Lead: Mansitha & Hirusha · SF personalisation likely sequential after #27, dev ending ~7 Aug latest
Analytics dashboard may link with future Avaya integration. RetailStudio (Maldives Transfers) excluded from scope.
⚠️ Key Dependencies & Risks
▼- 🔴 Azure resources — CHML to provision budget & create resources (target 24-Mar)
- 🔴 360SMS — Endpoint mapping & provisioning (target 24-Mar), dev brief by 27-Mar
- 🟠 Sabre SynXis — PO raised 17-Mar, sandbox target 27-Mar (via Antarish 20-Mar)
- 🟠 RedDot — API meeting by 20-Mar, going live on Brand.com 1-Apr
- 🟠 GHA — Clarification meeting by 19-Mar, need Sabre documentation
- 🟡 VAPT — Third-party security testing (~4 weeks), vendor TBC
- ⚪ RetailStudio (Maldives Transfers) — Excluded from scope
- ⚪ Salesforce Journeys — Deferred from MVP launch
👥 Project Team
▼| Name | Role | Key Responsibilities |
|---|---|---|
| Sayuni | Project Manager | Sign-offs, UAT coordination, testing, timeline management |
| Pamith | Developer | CMS knowledge retrieval, Google Maps, call centre info |
| Mansitha | Developer | Booking linkout, WhatsApp 360SMS, Sabre/RedDot/GHA integration |
| Hirusha | Developer | EatApp, spa linkout, analytics, cloud config, production migration, Sabre cert |
| Ransika / Naduni | Design & Dev | Canvas API menus, WhatsApp frontend design |
| CHML | Client / Owner | Azure budget, UAT, website integration, feedback, sign-off |
| Octave | Dev Partner | Integration support, dev briefs, analytics session data |
Salesforce Marketing Cloud
Internally managed · Journey Builder · Data Cloud activation
Overview: SFMC is managed by the internal team and serves as the primary execution layer for CRM journeys and email campaigns. The strategic roadmap centres on maturing Journey Builder automations and deeply integrating with Salesforce Data Cloud for audience activation.
📌 Current State
▼Ownership
Internal Team
SFMC operated in-house by the MarTech / digital marketing team
Connected Records
Consented Guests
Connected to consented guest records for email communications
Current Segmentation
Broad
Today: "all past guests," "all locals." Goal: granular, data-driven segments via Data Cloud
🔄 Journey Builder & Automation Flows
▼The roadmap involves moving from basic broadcast-style sends to sophisticated, trigger-based multi-step journeys driven by guest behaviour and lifecycle signals from Data Cloud segments.
Target State
Lifecycle Journeys
Trigger-based multi-step journeys: pre-arrival, in-stay engagement, post-stay winback, loyalty nurture, lapsed re-engagement
Activation Channels
Email + Ads
Email via SFMC; first-party audience segments pushed to Google & Meta for paid media
Personalisation
Data-Driven
Content personalisation via guest attributes, stay history, and segment membership from Data Cloud
🔗 Data Cloud Integration & Audience Activation
▼Current — SFMC + Consented Records
Now
SFMC connected to consented guest records. Broad segments. Data Cloud licensed but not activated for segmentation.
Near-Term — Data Cloud Activation
Phase 1 (2026)
Progress Data Cloud to operational use. Ingest Azure Synapse + Opera Cloud guest data. Build unified profiles and granular segments. Connect to SFMC Journey Builder and ad platforms.
Future — Full Closed Loop
Phase 2–3
EatApp F&B data integration, automated audience refresh to ad platforms, agentic automation feeding insights back into segmentation. AI Concierge WhatsApp personalisation via SF data retrieval (V1.5).
Cross-initiative link: Data Cloud is the shared backbone connecting Guest Analytics (Initiative 03), Agentic Audience Segmentation (Area 05), and AI Concierge personalisation (V1.5). Progressing it from early-stage to operational is a critical-path dependency across all four initiatives.
⚠️ Dependencies & Open Items
▼- 🔴 Data Cloud needs to move from licensed to operational — critical path for all segmentation work
- 🟠 Azure Synapse data pipeline into Data Cloud needs to be established
- 🟠 Opera Cloud OHIP guest data ingestion into Data Cloud
- 🟡 EatApp API connection — deferred to Phase 2 but needed for F&B behavioural segments
- 🟡 SF Journeys integration with AI Concierge WhatsApp — deferred from MVP to avoid downtime
- ⚪ Agentic automation integration — Phase 3 for closed-loop campaign intelligence
Guest Analytics & Campaign Intelligence
Data-driven segmentation · Salesforce Data Cloud · Replacing Octave's Guest 360
Vision: Replicate and expand Octave's Guest 360 analytics internally. Move from broad segmentation ("all past guests," "all locals") to granular, data-driven segments powered by Salesforce Data Cloud as the activation hub and Azure Synapse as the source of truth.
📄 Key Deliverables
▼Strategy Document
V2 · ~24 Pages
Segmentation framework, campaign activation architecture, agentic integration, phased roadmap, investment summary.
Architecture Diagram
Interactive SVG
React-based interactive architecture diagram, iterated across multiple revisions.
Naming Update
Cinnamon Discovery
"GHA Loyalty" renamed to "Cinnamon Discovery" and updated to Phase 1 in architecture diagram.
Budget
Within Ceiling
Phase 1 scoped within Year 1 budget. EatApp, automated ad activation, agentic integration deferred to Phases 2–3.
🏗 Architecture Components
▼| Layer | Component | Role | Phase |
|---|---|---|---|
| Data Source | Azure Synapse | Data lake & source of truth | Phase 1 |
| Data Source | Opera Cloud OHIP | PMS guest data via REST API | Phase 1 |
| Loyalty | Cinnamon Discovery | Loyalty programme integration | Phase 1 |
| Segmentation | Salesforce Data Cloud | Segmentation & activation hub | Phase 1 |
| Activation | Salesforce Marketing Cloud | Email / journey execution (consented records) | Phase 1 |
| Paid Media | Google & Meta Ads | Receiving first-party audience lists | Phase 1 |
| F&B Data | EatApp | Symphony POS aggregation via API | Phase 2 |
| Automation | Agentic Integration | Connecting to agentic automation areas | Phase 3 |
🗓 Phased Roadmap
▼Phase 1 — Foundation
2026
Azure Synapse + Opera Cloud OHIP data pipeline, Salesforce Data Cloud segmentation, Cinnamon Discovery integration, first-party audience activation to Google & Meta.
Phase 2 — Expand Data Sources
Future
EatApp API connection (F&B/Symphony POS), automated ad audience activation, expanded segmentation models.
Phase 3 — Agentic Integration
Future
Full integration with agentic automation for closed-loop campaign intelligence.
Open item: GHA references throughout the ~24-page strategy document still need updating to "Cinnamon Discovery." Diagram is updated but document is not — currently unresolved.
Agentic AI Automation
10 performance marketing agents · Internal ownership · Low-code/no-code first
Vision: Build an in-house agentic automation solution across 10 performance marketing task areas, eliminating dependency on external developers. The internal team defines segments and strategy; the performance agency builds and executes campaigns.
🔧 Technology Stack
▼
Power Automate PremiumAzure OpenAI GPT-4oAzure FunctionsClaude Max + ProAzure SynapseOpera Cloud OHIPSalesforce Data CloudSFMCGoogle AdsMeta AdsBenchDirect ProFornova RITripAdvisor BATHNMicrosoft 365Google Sheets
Key decision: Sabre and Synxis removed from architecture entirely. Azure Synapse + Opera Cloud OHIP is the confirmed data source. THN has no public admin API for pop-ups; TripAdvisor API limited to ratings/reviews.
📋 The 10 Automation Areas
▼| # | Area | Description | Feasibility |
|---|---|---|---|
| 01 | Tracker Updates | Auto-populate performance trackers (Google Sheets & SharePoint/Excel) with key metrics | High |
| 02 | Anomaly Detection & Alerts | Monitor campaign metrics, alert on significant deviations | High |
| 03 | Budget Pacing & Reallocation | Track spend vs budget, recommend reallocations | High |
| 04 | Reporting & Insight Generation | Automated performance reports with AI narrative insights | High |
| 05 | Audience Segmentation & Activation | Data-driven guest segmentation via SF Data Cloud, push to ad platforms | High |
| 06 | Bid & Budget Optimisation | AI-recommended bid adjustments and budget shifts | Moderate |
| 07 | Creative Performance Analysis | Analyse ad creative performance, recommend iterations | Moderate |
| 08 | Competitor Rate & Market Intel | Pull competitive data from BenchDirect, Fornova, TripAdvisor | Moderate |
| 09 | Campaign Launch Automation | Streamline end-to-end campaign setup and launch workflow | Moderate |
| 10 | Cross-Channel Attribution | Connect touchpoints across channels for journey & attribution insights | Lower |
🎯 Area 01 — Tracker Updates (First Prototype)
▼Metrics Confirmed
Revenue · ADR · Occupancy % · Conversion Rate · ROAS · Cost/Spend · OTB vs STLY
Tracker Locations
2 Platforms
Google Sheets + SharePoint / Excel Online
Update Cadence
Mixed
Some trackers daily, others weekly
Scoping Status
In Progress
Metrics & locations confirmed. Field mapping and automation flow design in progress.
🗓 Implementation Roadmap — Apr to Dec 2026
▼Phase 1 — Foundation & First Agents
Q2 2026 · April – June
Power Automate Premium environment, Azure OpenAI integration, Azure Functions scaffolding. Build Area 01 (Tracker Updates) as first prototype. Target: 3–4 high-feasibility agents live.
Phase 2 — Expand & Optimise
Q3 2026 · July – September
Expand to moderate-feasibility agents. Refine Phase 1 agents. Target: 3–4 additional agents live.
Phase 3 — Complete & Scale
Q4 2026 · October – December
Complete remaining agents incl. Cross-Channel Attribution. Full optimisation pass. Target: all 10 agents live by year end.
Investment: Architecture and investment plan finalised (CAPEX + OPEX + Year 1 total). Phase 1 within approved budget ceiling. Executive pitch drafted for CCO in detailed and summary variants.