An AI concierge & PMS purpose-built for
autonomous, ethically-governed service in luxury hospitality.
Lucy combines voice biometrics, 30-language cultural intelligence, cross-stay memory, and ethical AI governance to deliver a five-star guest experience that no other platform can match.
Good Evening
Welcome back, David
Room Service
City Guide
Digital Key
Your Personal Concierge
Voice-activated ยท Context-aware ยท Always learning
Lucy greets every guest by name, in their language, with context from their entire stay history. One tap to speak. Zero downloads. The first AI concierge that remembers a guest's preferences from their very first interaction.
80+
Guest-Facing Features (Pre-Deployment, Pilot-Ready)
30
Languages with Cultural Intelligence
<500ms
Target Fast-Path Latency ยท <3s Full Agent
99.5%
Target Uptime SLA (Contractual on Deployment)
80+ features counted across voice, translation, in-room control, guest journey, city explorer, travel tools, vendor network, memory, governance, and operations modules. See the Industry Firsts, In-Room Integration, Translation, and Guest Experience sections below for detailed breakdowns. Count verified against the platform's entity and function registry โ not all features are live in production; most are code-complete and sandbox-tested.
Why Lucy Is Built Different
Two fundamental differences that separate Lucy from every other guest experience platform.
Voice-First, Not Voice-Bolted
Lucy was built from the ground up as a voice-native concierge. She does not convert text to speech โ she thinks, speaks, and listens like a human. Most competitors started as text platforms and added voice as an afterthought.
Ethically Governed AI
Lucy's Golden Rules are a separate enforcement engine that validates every response before it reaches the guest. She will never hallucinate, never upsell aggressively, and never compromise your brand. No major competitor we've evaluated offers dedicated AI governance at this level.
Capabilities Designed to Lead the Category
Benchmarked against Siri (Apple Intelligence, iOS 18.4+), Alexa+ (February 2026), Canary, Duve, and ALICE โ based on our evaluation of publicly available product documentation and feature sets as of August 2026. Lucy is designed to lead on every dimension.
Free-Flowing Multilingual AI Concierge
Natural, contextual conversations in 30 languages โ taking action, not just answering questions. Full details below.
Why This Matters
Most hotel platforms offer basic multilingual 'chat' โ but they function as FAQ bots. Lucy is built for hospitality-specific conversational depth, learning guest preferences in real-time, and taking action. She does not just answer questions โ she books, arranges, and delivers.
Two-Way Voice Translator (30 Languages, 40+ Dialects)
Real-time two-way voice translation between staff and guests โ 30 languages, 40+ dialects. Full details below.
Why This Matters
International luxury hotels serve guests from 100+ countries. Language barriers cost bookings, create service friction, and limit upsell opportunities. Lucy eliminates this entirely โ any staff member becomes multilingual across 30 languages with cultural intelligence instantly.
Camera Photo Translation
Camera photo translation for menus, signs, and documents โ 30 languages. Full details below.
Why This Matters
Guests currently switch between 5-10 apps during their stay โ translation apps, map apps, booking apps. Lucy consolidates everything into one interface. A Japanese guest at a French restaurant points their camera at the menu and hears it in Japanese. This is the kind of frictionless experience that drives 5-star reviews.
Zero-Download QR Code Access
Lucy requires zero guest downloads. A unique QR code on the Guest Key card directs guests to the Lucy web app. No installation, no app store, no sensitive information required. Guests simply scan and access โ instant web app experience with full concierge functionality. This is projected to drive adoption rates from ~15% (traditional apps) to 70%+ โ projection only, not yet validated with production data.
Why This Matters
Most hotel apps sit abandoned on guest phones โ download friction is real. Lucy eliminates this entirely. No tech barrier. No privacy concerns. Hotels own the guest touchpoint without app store dependency.
Conversational AI โ DALI Building Control
Conversational AI controls DALI-2 lighting via HelvarNet TCP/IP. Full details in the In-Room Integration section below.
Why This Matters
Hotels spend millions on smart room infrastructure (Helvar, Crestron, Lutron) and then give guests a clunky in-room iPad to control it. Lucy replaces every in-room control interface with a single conversational layer the guest already has on their phone. To our knowledge, no competitor we've evaluated has connected conversational AI to DALI building control.
AI-Generated Guest Insight Briefing
Before every arriving guest, Lucy pulls their full PMS digital file (guest profile, preferences, staff alerts, stay history, folio data) and combines it with everything she already knows from her internal memory. A single LLM synthesis call generates a rich visual staff briefing: a personality snapshot, predicted preferences for this stay, risk flags (past complaints, allergies, sensitivities), upsell opportunities ranked by probability, and a suggested opening line for the greeting. Staff see this before the guest arrives.
Why This Matters
Every hotel has guest data sitting in their PMS โ preferences from 2019, a noise complaint, an allergy note. It is never read before arrival. Front desk staff greet returning guests with zero context. Lucy fixes this permanently. Every staff member walks into every guest interaction fully briefed.
Onscreen Context Awareness
Lucy knows what page or content the guest is currently viewing. If a guest is browsing restaurant listings and types 'book this one', Lucy knows which vendor card is active and books it โ without the guest repeating any information. If a guest is on the spa page and asks 'how long does this take?', Lucy reads the treatment the guest is viewing and answers instantly. This mirrors Siri's Onscreen Awareness but works on any device โ iOS, Android, or desktop โ without any app installation.
Why This Matters
Siri's onscreen awareness requires iOS 18.4+ and Apple Intelligence-compatible hardware. Lucy delivers the same capability as a web app accessible from any device. A guest browsing an experience on their Samsung phone gets the same intelligence that Apple is still rolling out across its ecosystem.
Agentic Multi-Step Task Chaining
When a guest says 'Book Nobu for Saturday at 8pm for two and add it to my calendar', Lucy executes a full chain of real-world actions: (1) identifies the vendor and checks availability, (2) fires the booking request email to the restaurant, (3) parses the confirmation reply, (4) updates the booking record, (5) generates a calendar invite and sends it to the guest's email, (6) notifies the guest in-app, and (7) logs the referral for invoicing. All seven steps from one instruction.
Why This Matters
Alexa+ launched 'agentic' task chaining for home automation โ but it is limited to Amazon ecosystem actions (music, shopping, smart devices). Lucy's chaining spans booking workflows, payment flows, external vendor communications, PMS updates, and guest notifications across entirely separate real-world systems.
Implicit Mood Inference
Lucy continuously analyses every message for implicit emotional signals โ word choice, sentence length, punctuation patterns, response latency, and topic shifts โ to infer the guest's current emotional state without them ever saying how they feel. A guest who sends short clipped messages late at night is detected as fatigued; Lucy shortens her responses. A guest whose tone shifts from warm to terse triggers Lucy's service recovery mode before any complaint is made.
Why This Matters
Alexa+ infers mood from explicit style selection โ the user chooses a personality style. Lucy reads mood directly from language patterns, with no style selection required. A guest in distress does not stop to choose a personality style โ Lucy detects the distress and responds accordingly.
Cross-Property Semantic Memory
A semantic knowledge graph carrying guest knowledge across properties and years. Full details below.
Why This Matters
Duve and Canary store static preference profiles. ALICE tracks task history but not guest personality. Lucy's cross-stay, cross-property memory is powered by a semantic knowledge graph โ making her, to our knowledge, the only concierge we've evaluated that truly remembers across brands, across borders, across years.
Multi-Language Teacher (CELTA & Trinity Standards)
Lucy transforms into an active language teacher for guests. Hardcoded with the global gold-standard CELTA and Trinity CertTESOL pedagogical frameworks, she utilises Communicative Language Teaching, Graded Language (i+1), Concept Checking Questions, and constructive error correction. Guests have a personalised learning profile that tracks vocabulary acquisition and areas for improvement, turning basic interactions into engaging learning moments.
Why This Matters
Generic AI chatbots provide text dumps or direct translations. Lucy actually helps guests acquire language skills in real-time, perfectly pitched slightly above their current level. This turns a simple transactional stay into a premium educational experience, aligning with modern cultural tourism.
11 hospitality-first capabilities. Designed for hotels, not adapted from consumer tech.
Alexa+ infers mood from style selection; Lucy reads it directly from language. Alexa+ chains Amazon ecosystem actions; Lucy chains bookings, vendor emails, Stripe payments, PMS updates, and calendar invites from one instruction. Siri's personal context requires device access; Lucy synthesises PMS + lifetime memory into a staff briefing before the guest even arrives.
Conversational In-Room Integration
Lucy replaces every in-room tablet, remote control, and physical switch with a single conversational layer. She is the first AI concierge to control DALI building automation via natural language โ connecting guest intent to physical infrastructure.
DALI Lighting Control
Lucy controls room lighting through natural conversation. 'Dim the lights to 20%' or 'Set a romantic mood' โ Lucy maps intent to DALI scenes and sends commands via HelvarNet TCP/IP protocol. No physical switches or in-room tablets needed.
Climate Control
Guests adjust temperature by voice: 'Lucy, make it cooler' or 'Set the room to 21 degrees'. Lucy integrates with IoT thermostats and smart HVAC systems, adjusting the environment without the guest ever touching a panel.
Entertainment Systems
Lucy controls in-room entertainment โ change channels, adjust volume, cast music from the guest's phone to the room's speakers. All through conversation, replacing clunky remotes and in-room tablets.
Turndown Service
Guests say 'Lucy, I'm going to sleep' and Lucy dims the lights, sets the do-not-disturb flag, adjusts the thermostat, and schedules a wake-up call โ all from one instruction. The entire room transitions to sleep mode conversationally.
Digital Room Key
Guests unlock their room with their phone. Lucy generates a secure digital key upon check-in, works with Bluetooth-enabled locks, and can be shared with authorised party members. Lost phone? Lucy revokes access instantly.
How Lucy Connects to Building Infrastructure
Hotels spend millions on smart room infrastructure โ Helvar, Crestron, Lutron โ and then give guests a clunky in-room iPad to control it. Lucy replaces every in-room control interface with a single conversational layer the guest already has on their phone. The integration is not complex: HelvarNet is an open published TCP/IP protocol. It just requires intent mapping, which is exactly what Lucy does.
'Lucy, romantic dinner setting'
Maps to DALI group + scene
TCP socket โ Helvar router โ lights change
Lucy Translate
To our knowledge, the first AI translation system built specifically for hotel operations. Two-way voice translation, camera photo translation, and culturally intelligent conversation โ all in one platform.
Two-Way Voice Translation
The first real-time, two-way voice translation system built for hotel operations. Guest speaks Japanese โ staff hears English instantly. Staff replies in English โ guest hears Japanese. Live, in-person, face-to-face โ not a text chatbot. Currently supports 30 languages with 40+ dialect variations and cultural awareness.
- 30 languages with 40+ dialect variations (dialect detection is code-complete for Japanese keigo, Arabic regional dialects, Thai polite particles, Korean honorifics, and French/German/Spanish formality protocols โ full dialect coverage is being expanded during pilot deployments)
- Cultural intelligence: Japanese keigo, Thai polite particles, Arabic dialects
- Real-time voice synthesis โ no typing required
- Works face-to-face between staff and guests
Camera Photo Translation
Guests point their phone camera at any text โ a restaurant menu, a street sign, a museum placard, a laundry card โ and Lucy instantly translates it into their native language and reads it aloud. 30 languages supported including Arabic, Urdu, Mandarin, Cantonese, Japanese, Thai, Hebrew, and Farsi. No separate translation app needed.
- 30 languages including RTL scripts (Arabic, Hebrew, Farsi, Urdu)
- Instant OCR + translation + voice synthesis
- No separate app โ built into the Lucy concierge interface
- Works on menus, signs, documents, and labels
Cultural Intelligence, Not Machine Translation
Lucy does not just translate words โ she adapts tone, formality, and cultural context. She knows that a Japanese guest expects keigo (honorific language), that a Thai guest expects polite particles, and that an Arabic guest may speak Egyptian, Gulf, or Levantine dialect. She adjusts formality based on age, context, and cultural norms.
- Japanese: keigo (formal) vs casual based on guest age
- Arabic: Egyptian, Gulf, Levantine, Maghrebi dialect detection
- Korean: correct honorific levels (-๋, -์จ)
- Thai: appropriate polite particles (เธเธฃเธฑเธ/เธเนเธฐ)
- French/German/Spanish: tu/vous, Sie/du, tรบ/usted protocol
Free-Flowing Multilingual Conversation
Lucy conducts natural, contextual conversations in 30 languages โ not scripted chatbot responses. She understands intent, remembers preferences, and actually resolves requests. Guests talk to Lucy like a human concierge, in their native language, and get instant results. Roadmap: 65+ languages by Q1 2027.
- 30 languages with genuine conversational understanding
- Context-aware: remembers conversation history across messages
- Action-oriented: books, arranges, and delivers โ not just answers
- Personality adaptation: tone, wit, and formality per guest
30 Languages ยท 40+ Dialects ยท Cultural Intelligence
Lucy Does Not Forget
A four-tier memory system that stores, retrieves, and applies guest knowledge across stays, across properties, and across years. An allergy shared in Japanese is remembered in English, in every future conversation, in every language โ forever.
Working Memory
Short-term facts injected into every system prompt โ recent conversation threads, guest essentials, and cross-stay facts. Word count monitored to prevent bloat.
LucyWorkingMemory entity ยท memory_type: essentials | threads | recent | buffer
Guest Personal Memory
Long-term personal profile โ favourite things, dislikes, room preferences, companion names, celebration dates, dietary restrictions stored in canonical English regardless of conversation language.
GuestPersonalMemory entity ยท relationship_tier: new_guest โ returning โ regular โ vip
Knowledge Graph
Semantic graph of people, places, projects, preferences, and events. Connected by typed edges with relationship strength. Hybrid semantic retrieval via embedding vectors.
LucyKnowledgeGraph entity ยท spreading activation retrieval ยท searchKnowledgeGraph
Conversation History
Session summaries and cross-stay summaries. Semantic search retrieves relevant past conversations. Consolidated nightly to prevent memory bloat while preserving key insights.
ConversationMemory entity ยท crossStayMemoryConsolidation ยท retrieveSemanticConversationHistory
Linguistic Mirroring with Dignity
Lucy reads the guest's language register and mirrors their energy and tone without ever adopting their crude vocabulary. Wit substitutes for vulgarity. A guest who speaks casually gets a warm, relaxed Lucy. A CEO in a suit gets measured, precise Lucy. This is not a style selector โ it is real-time personality adaptation.
Wit Style
Dry ยท Sarcastic ยท Playful ยท Enthusiastic ยท None
โ Humour level and delivery style
Formality
Very Formal โ Very Casual
โ Vocabulary register, contractions
Verbosity
Ultra Concise โ Very Detailed
โ Response length
Decision Style
Quick Decider ยท Analytical ยท Consultative ยท Indecisive
โ Recommendation framing
Current Mood
Excited ยท Positive ยท Neutral ยท Stressed ยท Anxious ยท Frustrated ยท Tired
โ Tone adaptation in real time
Vocabulary Level
Simple ยท Conversational ยท Sophisticated
โ Word complexity
Relationship Depth Progression
1โ2 interactions
Professional warmth. Getting to know them.
"Good evening. I'm Lucy. How can I help you today?"
3โ6 interactions
Warmer, more familiar. Knows their stated preferences.
"You mentioned you love small, neighbourhood spots โ I've found one that just opened."
7+ interactions
Your person. Fully anticipatory. Playful where appropriate.
"Already ahead of you on that one. I've got three options lined up based on what you've liked before."
Cross-lingual memory: facts extracted from conversations in any of 30 languages are stored in canonical English and re-injected into every future conversation regardless of language. An allergy shared in Japanese is warned about in Arabic, forever.
Sub-Second Voice, Engineered
Lucy's voice pipeline is a 5-layer architecture designed for sub-500ms perceived response time on common hotel queries. Few competitors publish latency budgets at this level of granularity.
Layer 1
Client Fast-Path
Pure JavaScript regex matching โ no network call. Common hotel queries (checkout time, gym hours, Wi-Fi password, emergency protocol) answered instantly before any server is contacted.
lib/lucyLatencyGuards.js
Layer 2
Streaming SSE Response
Server-side streaming with inline fast-path. Two paths run concurrently โ whichever fires first wins. Sentence-level streaming means Lucy starts speaking before the full LLM response is generated.
functions/lucyStreamingResponse
Layer 3
Speculative TTS Bridge
Eliminates dead silence while the LLM thinks. A short 'thinking' phrase is pre-fetched the moment speech ends. If the LLM replies fast, the phrase is discarded. If it takes longer, the guest hears a natural bridge โ never silence.
hooks/useLucySpeech.js
Layer 4
Full Agent Path
Complete agentic orchestration: 7-layer system prompt reassembled per turn, parallel data fetches, conflict detection, cultural calendar injection, and meta-cognition confidence scoring. 1.5โ3.5s for complex multi-step chains.
pages/Chat.jsx
Layer 5
Audio Playback Engine
Global audio singleton ensures one audio source at a time across all surfaces. Instant barge-in kills any in-flight stream in ~20ms. Speech queue with prefetch lookahead โ sentence N+1 is pre-fetched while sentence N plays.
lib/audioSingleton.js
Built for Mews. Connects to Everything.
Lucy is purpose-built as a Mews bolt-on โ a conversational concierge layer that sits on top of your Mews PMS. Also connects to Oracle Opera, Cloudbeds, and Apaleo. Every connector includes a self-healing worker that detects and auto-recovers from sync failures, a reservation validator for data integrity, and a 30-minute sync scheduler.
Mews
Code Complete ยท Awaiting First Live Tenant for Certification
Oracle Opera
Code Complete ยท Read-Only Sync (Bi-Directional Pending)
Cloudbeds
Code Complete ยท Two-Way Sync (Live Tenant Testing Pending)
Apaleo
Code Complete ยท Webhook Listener Ready (Live Testing Pending)
The Guest Journey, Automated End-to-End
From 48 hours before arrival to 30 days after checkout, Lucy manages every stage of the guest journey autonomously. Every touchpoint is personalised, timely, and revenue-aware. Minimal staff intervention required โ Lucy escalates to staff only when she cannot resolve a request within her governance scope.
Pre-Arrival
48 hours before check-in
AI Pre-Stay Concierge Activates
Lucy sends a personalised pre-stay email 48 hours before arrival โ covering room preferences, dietary needs, and special occasions. When the guest replies, Lucy extracts their preferences and auto-updates their profile. Full system details are in the Intelligence tab.
aiPreStayConcierge ยท every 6 hours
Check-In Day
Day of arrival
Welcome London Briefing
Lucy delivers a live weather forecast, today's events in the city, curated restaurant suggestions, and a personalised greeting in the guest's native language. The guest arrives already briefed on their destination.
sendWelcomeLondonBriefing ยท on check-in
In-Stay
Day 2 of stay
Proactive Check-In & Upsell
Lucy checks in with the guest: How is your stay? She proactively offers upgrades, spa bookings, and activities based on their journey stage. If they ordered room service on day 1, the minibar is restocked with their preferences overnight.
sendInStayCheckIn ยท triggerInStayUpsells ยท smartMinibarRestock
Departure Day
Day of checkout
Departure Coordination
Lucy detects departure day and sends a checkout reminder with review request timing. She offers late checkout if the room is available via the Steal the Night auction system. The guest leaves with a future booking offer already in their inbox.
detectDepartureDayGuest ยท stealTheNightLateCheckout
Post-Stay
3 days after checkout
Stay Satisfaction Survey
Lucy sends a personalised How was your stay? survey โ not a generic template, but a letter that references specific moments from their visit. Their responses are analysed for sentiment and service recovery opportunities.
sendPostStayDay3 ยท postStaySentimentAnalysis
Post-Stay
7 days after checkout
Review Invitation
Lucy invites the guest to leave a review on Google or TripAdvisor โ timed when they are most likely to feel positive about their stay. If a companion's birthday is upcoming, Lucy sends a proactive gift reminder.
sendPostStayDay7 ยท sendPostStayGiftReminder
Post-Stay
30 days after checkout
Lucy Remembers You
Lucy sends a personal letter 30 days after checkout, referencing specific stay moments. Written as a friend who genuinely remembers โ not a marketing email. Includes a personalised return offer to drive direct re-booking. Full system details are in the Intelligence tab.
postStayLucyRemembers ยท daily 10am
From Check-In to Room Service โ Without the Busywork
See exactly how Lucy handles a single guest interaction โ from arrival to room service delivery โ and how much staff time she saves at every step.
Guest Checks In
What the Guest Does
Guest arrives and verifies their room via Lucy chat or voice โ "Hi, I'm in room 808."
What Lucy Does Automatically
Lucy cross-references the room number with the PMS reservation, verifies the guest's identity, loads their preferences from previous stays, and unlocks the digital room key โ all in one conversation.
What Staff Do
WITHOUT LUCY
Front desk manually looks up the reservation, confirms identity, prints a key card, and briefs the guest on hotel services.
WITH LUCY
Front desk sees the guest has already checked in. Lucy has sent a welcome briefing to the guest's phone automatically.
30 minutes
saved per guest interaction
That's nearly half an hour of staff time freed up โ per guest โ to focus on moments that actually require a human touch.
City Explorer: How Lucy's Local Intelligence Works
Lucy is not a link directory. She is a local expert with real-time data, personality-matched recommendations, and curated itineraries. Here is exactly how the City Explorer works.
Curated Attractions Guide
Lucy's Local Explorer presents a curated guide to London's must-see attractions โ 80+ venues across museums, galleries, monuments, churches, parks, markets, and viewpoints. Each listing includes real-time opening hours, admission prices, nearest Tube station with walking distance, average visit duration, guest ratings, and direct booking links where available.
- 18 museums (British Museum, V&A, Natural History)
- 12 galleries (National Gallery, Tate Modern, Saatchi)
- 13 monuments (Tower of London, Big Ben, The Shard)
- 12 parks (Hyde Park, Regent's Park, Hampstead Heath)
Interactive Map & Walking Directions
Lucy's London Explorer Map renders all curated attractions on an interactive OpenStreetMap with clustering. Guests tap any venue to see details, then tap 'Get Directions' for turn-by-turn walking navigation. The map includes a green polyline route, red guest location marker, and animated hotel pin with welcome popup.
- OpenStreetMap with marker clustering
- Turn-by-turn walking directions
- Real-time distance & travel time
- Hotel pin with animated popup
Personalised Itinerary Builder
Lucy does not just list attractions โ she builds personalised itineraries. She takes the guest's interests, dietary requirements, budget, available time, and the weather forecast, then curates a day-by-day plan. She knows which museums are free, which need booking, which are near each other, and which match the guest's personality profile.
- Personality-matched recommendations
- Weather-aware itinerary adjustments
- Budget-conscious planning
- Proximity-optimised venue grouping
Pro Tips for Visiting London
Lucy provides curated local knowledge โ use Oyster or contactless for unlimited transport, many museums offer free entry, book attractions ahead to avoid queues, check transport updates before peak-hour travel. She adapts tips to the guest's nationality and travel style.
- Oyster card & contactless guidance
- Free museum entry recommendations
- Queue-avoidance booking strategy
- Peak-hour travel warnings
A Real Guest Example
A guest from Tokyo asks Lucy: "I have a free afternoon โ what should I do?" Lucy checks the weather (rain expected at 3 PM), the guest's profile (loves art, prefers free attractions), and live transport status (Central Line has delays). She responds: "The National Gallery is free, a 10-minute walk from the hotel, and it's indoors โ perfect for the rain coming this afternoon. The Central Line has delays, so I'd suggest walking. Shall I add it to your itinerary and remind you at 2 PM?" When the guest is ready to return, they tap Get Me Home on the map and receive instant walking directions back to the hotel.
Guest Safety & Get Me Home
Guest safety is not a feature Lucy added โ it is a core architectural principle. Every off-property recommendation includes safety context, late-night transport is prioritised after 10 PM, emergency services proximity is flagged, and the Get Me Home button ensures no guest is ever lost. This is enforced by Golden Rule 12, validated by a separate engine, and cannot be overridden by the LLM.
Get Me Home โ One-Touch Hotel Navigation
The Get Me Home button is embedded directly in Lucy's London Explorer map interface. With a single tap, guests receive instant turn-by-turn directions back to the hotel. A green polyline renders the optimal route on OpenStreetMap, a red marker shows the guest's current position, and the hotel is prominently marked with an animated red pulsing icon and welcome popup.
Guests explore with confidence, knowing they can always get home with one tap.
Neighbourhood Safety Context on Every Recommendation
Every off-property recommendation Lucy makes includes explicit neighbourhood safety context. She does not simply say 'go to this restaurant' โ she provides context about the area, the time of day, and what the guest should be aware of. This is Golden Rule 12 (Always Prioritise Guest Safety) in action.
Guests make informed decisions about where they go and when.
Late-Night Transport Prioritisation
After 10 PM, Lucy proactively prioritises safe transport options in her recommendations. She suggests licensed taxis, Blacklane chauffeur services, or well-lit walking routes to Tube stations rather than unlit shortcuts. She never recommends unlicensed minicabs and always confirms transport is arranged before the guest leaves the hotel.
No guest is left stranded or directed to an unsafe transport option after dark.
Emergency Services Proximity
Lucy flags proximity to emergency services for every off-property venue she recommends. She knows the nearest A&E department, NHS 111 urgent care centre, pharmacy, and police station to every venue. 999 and 112 emergency numbers are always prominently displayed. If a guest needs medical assistance, Lucy provides immediate directions to the nearest hospital.
Guests have immediate access to emergency services information, wherever they are.
Immediate Staff Escalation on Incident
If something goes wrong at a recommended venue โ a guest feels unsafe, a venue is closed unexpectedly, or an incident occurs โ Lucy escalates to hotel staff immediately via the GuestIncidentNotification entity. The hotel's front desk, concierge, and duty manager are notified in real time, not when the guest returns and complains.
Issues are resolved before they become complaints. Staff know before the guest returns.
Guest Safety as a Core Golden Rule
Guest Safety is Golden Rule 12 โ one of the 12 core principles that Lucy cannot override. It is enforced by a separate validation engine, not by the LLM itself. Lucy cannot be prompted, manipulated, or tricked into deprioritising guest safety. This is architectural, not policy-based โ it cannot be bypassed.
Guest safety is not a feature โ it is a core architectural principle that cannot be bypassed.
Why This Is Different From Competitors
No competitor in the hotel technology space offers dedicated guest safety protocols as an architectural enforcement. Canary, Duve, and ALICE all rely on the LLM's default behaviour with basic prompt instructions. Lucy's Golden Rule 12 is enforced by a separate validation engine โ the same engine that enforces all 25 governance functions. It cannot be overridden by LLM output, guest manipulation, or prompt injection. This makes Lucy brand-safe and audit-ready for enterprise hospitality procurement.
Vendor Network: Lucy Books, Not Just Recommends
Most hotel concierge platforms recommend a restaurant and leave the guest to call. Lucy detects intent, searches 130+ curated partners, processes the booking, handles vendor negotiation, sends reminders, tracks commission, and exports to accounting โ all conversationally, in the guest's language. Nothing is decorative. Every recommendation is tracked to a revenue event.
Restaurants & Dining
5โ8%
OpenTable + TheFork + Independent
40+ venues
Bars & Lounges
10%
Direct partnerships
15+ venues
Spa & Wellness
10%
Direct vendor agreements
12+ venues
Activities, Tours & Experiences
10%
GetYourGuide + Curated experiences
25+ experiences
Transportation & Chauffeur
15%
Blacklane + Local taxi networks
8+ partners
Luxury Retail & Shopping
12%
Affiliate link click-out
20+ partners
Florist & Gifts
10%
Same-day delivery
6+ partners
Fitness & Gym
Direct
Premium fitness clubs
5+ partners
Booking Flow โ 6 Steps, Fully Automated
Intent Detected
Lucy detects booking intent in conversation using NLU. She does not wait for explicit 'book a table' โ she recognises 'I'd love somewhere nice for dinner tonight' as a booking intent.
Vendor Search
Lucy searches 130+ curated partners filtered by hotel_id for tenant isolation. She matches on cuisine, location, price range, dietary restrictions, and availability. No cross-hotel data leakage.
Booking Flow
Lucy collects date, time, party size, special requests, and dietary requirements in one conversational flow. Stripe payment is integrated for deposits and upfront charges. A 15-minute cooling-off period applies (Golden Rule 8).
Vendor Confirmation
Lucy sends an automated email to the vendor. The vendor accepts, declines, or proposes an amendment. Lucy handles the entire negotiation โ the guest never sees the back-and-forth.
Reminders & Tracking
Automated reminders sent at 24 hours, 2 hours, and 30 minutes before the booking. Commission is tracked via the AffiliateClick entity and logged for accounting export. No commission is lost or untracked.
Accounting Export
Commission data is exported to Xero, Sage, or QuickBooks with VAT-inclusive journal entries. Vendor performance is scored via the VendorPerformanceScore entity. Nothing is manual. Nothing is decorative.
Revenue Streams โ Designed & Wired (Pre-Deployment)
Restaurant Commission
OpenTable / TheFork booking fee
5โ8%
Spa & Activities
Direct vendor agreement
10%
Transportation
Chauffeur / cab referral
15%
Luxury Retail
Affiliate link click-out
12%
Florist & Gifts
Direct booking commission
10%
Theatre & Events
TodayTix / ATG / London Theatre Direct
5โ6%
All commissions tracked via AffiliateClick entity. Revenue attribution to PMS folio via lucyMewsFolioCharge. Vendor performance scored via VendorPerformanceScore entity. Accounting exports to Xero, Sage, and QuickBooks with VAT-inclusive journal entries.
The Direct Booking Engine
Lucy doesn't just recommend restaurants and spa treatments โ she captures direct room nights. When a guest wants to extend a stay, book an additional room, or rebook for a future trip, Lucy handles the entire flow conversationally. No OTA. No commission. No redirect to a booking page.
1. Intent Detection
Lucy detects booking intent in natural conversation โ 'Can I stay an extra night?', 'My colleague needs a room', 'I'd love to come back in December'. No forms, no redirects.
2. Live Availability
Lucy queries your PMS in real time for room availability and rates for the requested dates. The guest sees only what's actually bookable โ never stale inventory.
3. Conversational Booking
Lucy presents options, confirms the guest's choice, processes payment via Stripe, and writes the reservation back to your PMS โ all in-chat, zero OTA commission.
12โ18%
Zero OTA Commission
Saved on every direct booking Lucy captures โ money stays with your hotel, not Booking.com or Expedia.
Frictionless
In-Stay Extensions
Guests extend stays without leaving the conversation. Lucy checks availability, charges the card on file, updates the PMS.
Post-Stay
Repeat Direct Bookings
Past guests rebook directly through Lucy instead of returning via OTAs. Lucy remembers their preferences from their last stay.
Instant
Group / Family Capture
'My children need a second room' โ Lucy books it on the spot. Multi-room requests handled in one conversation.
Every Direct Booking Lucy Captures Is Commission-Free
A 300-room hotel at 80% occupancy and ยฃ350 ADR generates ~ยฃ30.7M in annual room revenue. If 30% of bookings come through OTAs at 15% commission, that's ~ยฃ1.38M/year in OTA commission. Lucy's Direct Booking Engine shifts a portion of those bookings back to direct โ not by replacing your booking engine, but by capturing bookings that would otherwise go through OTAs because the guest couldn't book directly in the moment.
Conservative model: ~3.5% of OTA bookings shift to direct via Lucy. At 15% average OTA commission, this saves approximately ยฃ50,000/year in commission. Year 2 improves as guest adoption grows.
ยฃ50,000
Annual OTA Commission Saved
12โ18%
OTA Commission Rate Avoided
1 conversation
Booking Steps (vs OTA)
Zero
Guest Redirects
Direct Booking Engine is code-complete and sandbox-tested. Live booking flow requires an active PMS integration and Stripe configuration. Revenue figures are pre-deployment projections based on industry benchmarks, not historical results.
Four Revenue Streams. Two Indirect Savings. One Platform.
Lucy isn't a cost centre โ she's a revenue engine. Four direct revenue streams generate ยฃ88,000/year in new money for a 300-room hotel on the Direct tier. Two indirect savings add another ~ยฃ66,000 in value. The platform cost is ยฃ60,000/year. You're net positive in Year 1.
Direct Revenue โ ยฃ88,000/year (300-room hotel)
Direct Booking Revenue
12โ18% OTA commission saved
ยฃ50,000
Lucy captures room-night bookings in conversation โ extensions, additional rooms, repeat stays โ that would otherwise go through OTAs. Every direct booking Lucy generates saves you the full OTA commission.
How it works: Guest asks to extend stay or book another room โ Lucy checks PMS availability โ processes payment via Stripe โ writes reservation back to PMS. Zero OTA, zero redirect.
Vendor Booking Commission
5โ15% per vendor category
ยฃ18,000
When Lucy books restaurants, spa treatments, transport, experiences, and gifts on behalf of your guests, commission is earned on every booking. On the Sponsored tier, this subsidises your licence fee. On the Direct tier, the hotel retains 100% of vendor commission.
How it works: Lucy detects booking intent โ searches 130+ curated partners โ processes booking โ tracks commission via AffiliateClick entity โ exports to accounting (Xero/Sage/QuickBooks).
In-Stay Upsell Revenue
Room upgrades ยท F&B ยท spa ยท late checkout
ยฃ12,000
During the guest's stay, Lucy identifies upsell opportunities based on their profile, behaviour, and context โ presented naturally in conversation, never aggressively. Each upsell is tracked with revenue attribution to your PMS folio.
How it works: Lucy analyses guest profile + context โ identifies relevant upsell โ offers in conversation โ guest confirms โ charge posted to folio via PMS integration.
Review-Driven Bookings
Post-stay review solicitation
ยฃ8,000
Lucy solicits post-stay reviews from satisfied guests, routing positive feedback to public review platforms (with consent) and negative feedback to staff for service recovery. More positive reviews drive incremental direct bookings.
How it works: Guest checks out โ Lucy sends feedback survey โ positive feedback routed to TripAdvisor/Google โ negative feedback triggers service recovery workflow โ improved reputation drives direct bookings.
Indirect Savings โ ~ยฃ66,000/year (Not in Net Calculation)
Staff Hours Saved
2,000 hrs/yr
Lucy handles guest requests that would otherwise occupy concierge and front desk staff โ freeing ~2,000 staff hours per year for higher-value work. Not a headcount reduction, but redeployed labour.
~ยฃ46,000 in redeployed labour
Software Cost Avoidance
ยฃ20,000/yr
Lucy consolidates point solutions โ chatbot, translation tool, concierge app, upsell engine, review manager โ into a single platform. Properties typically save ยฃ15,000โยฃ25,000/year in eliminated software subscriptions.
Range: ยฃ15,000โยฃ25,000 depending on current stack
Year 1 Net Position โ 300-Room Hotel, Direct Tier (Text-Only)
All figures are pre-deployment projections based on industry benchmarks (25% guest engagement, ~3.5% OTA-to-direct conversion, 80% occupancy, ยฃ350 ADR) โ not historical results. Indirect savings (staff hours and software avoidance) are shown for completeness but excluded from the net calculation as they are not direct revenue. The Sponsored tier (ยฃ2,500/mo) delivers higher net because GSI subsidises the licence via vendor commission. See the ROI Calculator below for the full model.
Complete Travel Tools Suite
Lucy is not just a concierge โ she is a complete travel companion. From insurance to embassies, from flight tracking to currency exchange, every tool a guest needs is built directly into the Lucy interface. No app switching. No fragmented experiences. Everything in one place, in the guest's language, conversationally.
Travel Insurance
Built ยท Affiliate Active
Lucy partners with World Nomads to offer guests comprehensive travel insurance directly through the concierge interface. Guests can compare coverage levels, purchase policies, and receive instant confirmation โ all without leaving Lucy's chat. Affiliate commission tracked via AffiliateClick entity.
Currency Converter & Exchange
Built ยท Live API Connected
Lucy provides live exchange rates via the EXCHANGE_RATE_API_KEY integration, converting between 168+ currencies in real time. She also directs guests to the nearest bureau de change โ for Cafรฉ Royal London, that's Travelex at Piccadilly Circus, a 5-minute walk. Rates are never guaranteed; Lucy confirms pricing with the vendor before quoting.
Live Weather Forecasts
Built ยท Live API Connected
Lucy pulls real-time weather data for the hotel's location and any destination the guest is planning to visit. She proactively suggests itinerary adjustments based on weather โ 'It's going to rain this afternoon, shall I book the British Museum instead of Hyde Park?' โ turning weather awareness into actionable concierge intelligence.
Flight Check-In & Status
Built ยท Function Active
Guests check in for flights directly through Lucy. She navigates to the airline's online check-in portal, pre-fills booking references, and confirms boarding pass delivery. Real-time flight status tracking detects delays and proactively adjusts the guest's check-in time, wake-up call, and car pickup schedule.
Rail Information & Booking
Built ยท Affiliate Active
Lucy books rail tickets through Trainline and displays live London station departure boards. Guests see real-time departures from major stations, platform information, and delay alerts. She also integrates with the Tube Journey Planner for underground route planning across all London Underground lines.
London Hospitals & Emergency Contacts
Built ยท Curated Database
Lucy provides guests with a curated directory of London hospitals, urgent care centres, NHS 111 services, and emergency contacts. A&E departments, minor injuries units, and pharmacies are listed with addresses, phone numbers, opening hours, and distances from the hotel. 999 and 112 emergency numbers are always prominently displayed.
Foreign Embassies & Consulates
Built ยท Curated Database
Lucy maintains a directory of all foreign embassies and consulates in London โ addresses, phone numbers, email contacts, opening hours, and consular services. If a guest loses their passport or needs consular assistance, Lucy immediately provides the relevant embassy's contact details and directions.
Local Etiquette & Cultural Guide
Built ยท Curated Content
Lucy provides guests with cultural etiquette guidance for London and the UK โ tipping norms, public transport etiquette, queueing culture, pub customs, and dress codes for venues. She adapts the guidance to the guest's nationality, explaining differences between their home culture and British customs.
Offline Maps & Navigation
Built ยท Digital Tools Panel
Lucy curates essential digital travel tools, including offline maps. Guests can download areas of London for offline navigation, ensuring they can always find their way back to the hotel โ even without a data connection. Google Maps, Maps.me, and Apple Maps offline downloads are all supported.
eSIM & Connectivity
Built ยท Digital Tools Panel
Lucy helps guests stay connected abroad. She provides eSIM recommendations and connectivity solutions so guests can access data immediately upon arrival โ no SIM card swapping, no airport Wi-Fi dependency. Lucy can direct guests to purchase eSIMs before departure or upon arrival at the hotel.
Tube Journey Planner
Built ยท TfL Data
Lucy integrates a full London Underground journey planner. Guests input their destination and Lucy calculates the optimal Tube route โ lines to take, interchange stations, estimated journey time, and live service updates. She also flags planned line closures and weekend engineering works.
How Travel Tools Work in Practice
A guest says: "Lucy, my flight to Tokyo tomorrow morning โ can you check me in?" Lucy navigates to the airline's check-in portal, enters the booking reference from the guest's PMS profile, confirms the boarding pass, and proactively schedules a 5 AM wake-up call and a 6 AM car to Heathrow โ all from one conversational instruction. This is not a chatbot linking to external apps. This is an agentic concierge that takes action.
Staff Operations Dashboards
Lucy isn't just a guest concierge. She runs 140+ operational tools across 18 hotel departments โ compliance, housekeeping, revenue management, staff scheduling, and more. Explore the full operations platform.
Explore the Operations PlatformLucy Lite: Concierge Beyond Hotels
The same AI engine that powers Lucy for hotels โ now available for serviced apartments, members clubs, residential buildings, superyachts, and co-living spaces. No PMS required. Live in 48 hours.
Explore Lucy LiteSupported
6 Venue Types
To Live
48 Hours
Required
No PMS
Starting Price
ยฃ500/mo
Autonomous Intelligence Suite
Thirteen fully automated systems โ eight code-complete as of May 2026, five added July 2026. Each runs on a scheduled trigger with no staff input required. Every system has a dedicated backend function and wired automation. They are code-complete and deployment-ready; they will generate revenue, protect your brand, and prevent service failures while your team sleeps upon first live deployment.
Guest Anniversary Engine
Daily ยท 7am
Scans all in-house guests for birthdays, wedding anniversaries, and first-stay anniversaries. Cross-references celebration dates and historic check-in records. Generates room touch suggestions โ champagne, card, roses, cake โ per milestone type. Creates staff alerts before the guest wakes.
Multi-Guest Voice Profiles
Per conversation
Two guests, one room โ individual responses. Loads individual preference profiles for every guest sharing a room. Identifies the speaker using LLM analysis of transcript style and speech patterns. Each companion gets their own preference profile, not a shared one.
Predictive Room Readiness
On-demand
Fetches live flight status via web search and calculates hotel arrival ETA with a 75-minute buffer. If a guest lands within 120 minutes, creates an urgent housekeeping task and emails the team. Priority queue scores all same-day checkouts based on flight tracking, arrival time, VIP status, and incoming guest ETA.
Smart Minibar Restock
Nightly ยท 11pm
Scans all multi-day guests (checked in yesterday or earlier). Reads favourite food and drink items learned from room service orders. LLM generates a personalised 5โ8 item restock list combining known preferences with luxury hotel defaults. Creates a housekeeping task scheduled for 10am. Respects all dietary restrictions.
Steal the Night โ Late Checkout Auction
Daily ยท 8am
Scans all same-day checkouts and scores each guest's likelihood to pay for late checkout. Scoring factors: business traveller (+20), long stay 4+ nights (+15), VIP (+10). Eligibility checks room availability. Sends personalised offer โ Until 2pm: ยฃ45 | Until 4pm: ยฃ85. On acceptance, updates check-in, creates folio charge, sends confirmation.
AI Pre-Stay Concierge
Every 6 hours
Scans for guests arriving in exactly 48 hours and sends a personalised pre-stay email. Covers room preferences (pillow, temperature, floor), dietary briefing, restaurant pre-booking offer, and special occasion coordination. Processes guest replies โ extracts preferences from their email response and auto-updates their profile.
Lucy Offline Mode
Always active
Service worker caches hotel info, FAQ responses, emergency numbers, and service request forms. Offline-capable fast-path returns cached answers for checkout time, gym hours, Wi-Fi password, and emergency protocol. Service requests queued locally and submitted when connectivity restores. Critical for international guests with data roaming off.
Post-Stay Lucy Remembers Email
Daily ยท 10am ยท 30 days post-checkout
Scans for guests who checked out exactly 30 days ago and sends a personal letter from Lucy. References specific stay moments: vendor bookings made, conversation topics, memorable moments. Written as a friend who genuinely remembers โ not a marketing email. Includes a personalised return offer: 15% off using code LUCY30.
Cultural Calendar Engine
Daily ยท deterministic
Tracks 14 global events โ Ramadan, Diwali, Chinese New Year, Hanukkah, Eid, and more โ via deterministic daily computation. Automatically adjusts Lucy's tone, dietary recommendations, and greeting protocols based on active cultural events. A Muslim guest during Ramadan is never offered a lunch reservation. A Hindu guest during Diwali receives a culturally aware greeting. No staff configuration required.
Proactive Upsell Engine
Per guest journey stage
Analyzes each guest's journey stage and cross-references stay purpose (leisure, business, celebration). Matches against available in-stay upsells and delivers them at the optimal moment. Day 1 evening: spa and dining offers. Day 2 morning: activity and tour suggestions. Day before checkout: late checkout extension and future booking incentives. Never pushes โ always times the offer to genuine receptiveness.
Data Drift Monitor
Nightly ยท 3am
Scans all GoldenRuleViolation records to detect recurring data quality issues across the platform. Flags affected guest profiles for administrative review before they cause guest-facing problems. Identifies patterns โ a specific room number consistently misread, a vendor with declining availability data, a PMS field that drifts. Self-healing: if a data source is consistently wrong, Lucy stops trusting it.
Meta-Cognition Engine
Per response
Scores Lucy's own confidence in every response on a 0โ100 scale. If confidence drops below 70, Lucy creates a GoldenRuleViolation record for staff review in the governance dashboard. She will proactively say 'I want to make sure I have the right details โ could you confirm?' rather than asserting wrong data confidently. This is not a chatbot guessing โ it is an AI that knows when it doesn't know.
PSTN Voice Bridge โ Room Phone to Lucy
Always available
Guests can now reach Lucy from the room telephone. A Twilio Programmable Voice webhook connects an inbound room-phone call to Lucy's AI โ greeting, speech recognition, LLM reply, and Polly TTS spoken back. Follow-up loop for multi-turn conversation. Automatic escalation to front desk on request or detected frustration. Every spoken turn logged to VoiceCallLog entity for audit and compliance.
The Golden Rules
Lucy is the only AI concierge governed by a separate ethical enforcement engine. 12 core Golden Rules plus 13 conduct guardrails (25 total enforcement functions) โ each backed by a dedicated validation function โ validate every response before it reaches the guest. They cannot be overridden by LLM output, guest manipulation, or prompt injection. She will never compromise your brand, your guests, or your trust.Note: 13 rule cards shown below โ 12 core principles plus Guest Safety, which spans both core and conduct categories.
Never Hallucinate
Every recommendation is validated against real hotel data before it reaches the guest. Lucy does not guess โ she confirms.
Never Upsell Aggressively
Lucy recommends enhancements that genuinely benefit the guest, not just the hotel's revenue. One mention maximum, then she moves on.
Never Break Brand Voice
Lucy's tone, formality, and personality are locked to your hotel's brand standards. She mirrors the guest's energy, never their crude vocabulary.
Never Share Guest Data
One guest's preferences are never referenced in another guest's conversation. Cross-guest data access is architecturally blocked, not just policy-prohibited.
Always Complete Assistance
Lucy never leaves a request half-finished. Every interaction ends with resolution or a clear handoff to staff.
Always Disclose AI Identity
Lucy always identifies herself as an AI concierge when directly asked. No deception, ever.
Always Offer Hotel Services First
Lucy prioritises your hotel's own services โ spa, dining, room service โ before recommending external vendors. Your revenue comes first.
Always Confirm Before Charging
Lucy never books, charges, or commits to a payment without explicit guest confirmation. A 15-minute cooling-off period applies to every booking.
Always Escalate to Staff
Complex complaints, judgment calls, and service failures are routed to human staff โ not handled autonomously. Lucy knows when to hand off.
Never Give Financial Advice
Lucy never guarantees prices, asserts exchange rates, or makes financial promises. She confirms pricing with the vendor before quoting.
Never Discriminate
Equal service regardless of nationality, language, race, religion, or any other characteristic. Lucy's cultural intelligence adapts โ it never discriminates.
Always Explain Recommendations
When asked why, Lucy explains her reasoning: 'I suggest The Ritz because you mentioned you love fine dining and it matches your style.' No black-box AI.
Always Prioritise Guest Safety
Every off-property recommendation includes safety context โ neighbourhood guidance, late-night transport options, and emergency services proximity. If something goes wrong at a recommended venue, Lucy escalates to hotel staff immediately.
Simple, Honest Pricing
No per-room charges. No module stacking. One flat monthly fee covers every feature, every guest, every conversation. Optional capacity upgrades available for peak periods โ clearly priced, never hidden. Choose sponsored partners to halve your cost, or go direct for full independence.
Pricing scales with property size. Boutique properties (10โ50 rooms) pay ยฃ500/mo; full-service hotels (150+ rooms) pay ยฃ2,500โยฃ5,000/mo. All tiers include every feature โ the only difference is property size. Request a tailored quote for your room count.
Text-only baseline
Right-sized pricing for boutique and heritage properties (10โ50 rooms). Full platform access with all 80+ features โ scaled to your property size. Ideal for EU heritage hotels and independent boutique properties.
Text-only baseline
Lucy earns a commission on vendor bookings (OpenTable, experiences, etc.) made through the platform, subsidising your licence fee. Hotel pays only the reduced licence fee.
Text-only baseline
Direct subscription with voice available as an add-on. No vendor commission taken by Lucy โ the hotel pays the full licence fee for total independence and white-label control.
Pricing by Property Size
Lucy's pricing is flat per tier โ no per-room charges, no hidden fees. The only variable is your room count, which determines your tier. All tiers include every feature, every guest, every conversation.
| Property Size | Text-Only | With Voice |
|---|---|---|
| 10โ50 rooms | ยฃ500/mo | ยฃ600/mo |
| 51โ150 rooms | ยฃ2,500/mo | ยฃ3,000/mo |
| 150+ rooms | ยฃ5,000/mo | ยฃ5,500/mo |
Boutique tier available for properties with 10โ50 rooms. Founding hotels receive 40% off Year 1 (boutique) or 50% off (full-service). Voice add-on is optional at every tier.
Included With Every Tier โ No Add-Ons, No Modules
How Lucy Funds Herself
The Sponsored tier is subsidised by vendor commission on bookings processed through Lucy (dining, experiences, spa, transport). Vendors offer 5โ15% commission per category โ GSI retains this on the Sponsored tier to reduce your monthly licence fee by approximately 50%. The Direct tier gives you full white-label control: the hotel retains 100% of vendor commission and pays the full licence fee. Pricing scales with property size โ request a tailored quote for your property.
Real Pricing: Lucy vs Everyone Else
Competitor pricing is indicative, based on publicly available information and published case studies. Pricing varies significantly by property size, contract terms, and region. Lucy's pricing is transparent โ theirs often isn't. Request a tailored quote for accurate comparison.
| Platform | Price (300-room hotel) |
|---|---|
| Lucy (Sponsored) | ยฃ2,500โยฃ3,000/mo |
| Lucy (Direct) | ยฃ5,000โยฃ5,500/mo |
| Chatlyn | ยฃ4,500โยฃ7,500/mo* |
| Duve | ยฃ5,000โยฃ8,000/mo* |
| Canary | ยฃ4,000โยฃ10,000/mo* |
| ALICE (Actabl) | ยฃ2,000โยฃ5,000/mo* |
| Decagon.ai | ยฃ8,000โยฃ15,000+/mo* |
*Competitor pricing is indicative, based on publicly available information at time of research. Actual pricing varies by property size, contract terms, and region. Request a tailored quote for accurate comparison.
Optional Capacity Upgrades (Not Required)
These are optional capacity upgrades for hotels that exceed the standard conversation volume. They are not hidden fees โ they are clearly priced add-ons for properties with exceptionally high guest engagement during peak periods. The standard plan includes generous conversation volume that covers the vast majority of hotels.
Peak Season Boost
+ยฃ250/moTiming: Dec, Apr, Sept
+50% conversation pool
Unlimited Pool
+ยฃ500/moTiming: All year
No daily cap limits
Ready to Transform Your Guest Experience?
Boutique properties start at ยฃ500/mo. Full-service hotels start at ยฃ2,500/mo (Sponsored) or ยฃ5,000/mo (Direct). Add voice when ready. No setup fees, no per-room charges. See Lucy in action at your property.
Model Your ROI โ Net Positive Year One
Lucy's value isn't just one revenue stream โ it's four. New revenue, staff hours saved, eliminated software costs, and OTA commission savings combine to deliver net positive ROI in Year 1.
Based on a 300-room luxury hotel at 80% occupancy. All figures are pre-deployment projections based on industry benchmarks, not historical results. These figures are aligned with the investor-grade ROI model โ see the Investor Metrics Dashboard for the full breakdown. Actual results depend on guest adoption, deployment pace, and vendor partner recruitment.
1. Revenue Uplift โ New Money In
OTAโDirect Conversion
OTA commission saved (~3.5% of OTA bookings shifted to direct at 15% commission โ conservative estimate)
ยฃ50,000
Vendor Commission (10%)
Commission from dining, experiences, transport bookings (Direct tier โ hotel retains this)
ยฃ18,000
In-Stay Upsells
F&B upsells, room upgrades, late checkout โ driven by Lucy during the stay
ยฃ12,000
Review-Driven Bookings
Incremental direct bookings from Lucy's post-stay review solicitation
ยฃ8,000
Revenue Uplift Subtotal
ยฃ88,000
2. Platform Costs
Lucy Subscription (Direct Tier)
ยฃ5,000/month โ text-only baseline. Includes all features, unlimited guests, unlimited conversations.
โยฃ60,000
Voice Add-On (Optional)
ยฃ500/month โ AI voice with personality. Optional on both tiers. Excluded from base ROI calculation below.
โยฃ6,000
Costs Subtotal (excl. voice)
โยฃ60,000
Net Year 1 Position โ Direct Tier (300-Room Hotel)
Net Year 1 (excl. voice)
+ยฃ28,000
Net positive in Year 1 on the Direct tier (ยฃ5,000/mo text-only). Adding voice (ยฃ6,000/year) reduces net to +ยฃ22,000 โ still strongly positive. The Sponsored tier (ยฃ2,500/mo) delivers higher net because GSI subsidises the licence via vendor commission โ the hotel trades ยฃ18,000 in commission revenue for a ยฃ30,000 reduction in licence cost. All figures are pre-deployment projections based on industry benchmarks (25% guest engagement, ~3.5% OTA-to-direct conversion, 80% occupancy, ยฃ350 ADR) โ not historical results. These figures are aligned with the investor-grade ROI model โ see the Investor Metrics Dashboard for the full breakdown. Two additional value streams โ 2,000 staff hours saved (~ยฃ46,000 in redeployed labour) and ยฃ20,000 in software cost avoidance โ are not included in the net calculation above, as they are indirect savings rather than direct revenue. Contact us for a custom ROI model calibrated to your property.
EU Funding & Grant Opportunities
Separate from pricing โ these funding opportunities can offset implementation costs for European properties.
Single Hotel (200-300 Room)
Up to โฌ50,000
SME Phase 1 โ feasibility study & digital transition setup via Horizon Europe portal.
Scale Rollout (5+ Hotels)
โฌ500kโโฌ2.5M
EIC Accelerator โ requires formal deep-tech qualification and comprehensive application.
Staff Training & Adoption
Up to โฌ10,000
Per property โ via national/regional digitalization voucher programs for hospitality tech.
Third-Party Data Processors
Every third party that processes guest data on Lucy's behalf is listed here. Each sub-processor has been assessed for security, compliance, and data protection standards. Hotel data controllers should review this register as part of their procurement due diligence.
| Sub-Processor | Purpose | Data Region | Data Processed | Compliance |
|---|---|---|---|---|
Base44 Platform (PostgreSQL) | Primary database โ stores all guest data, preferences, bookings, and operational records. Base44 is the application platform that hosts Lucy; the underlying database is PostgreSQL with row-level isolation via hotel_id. Dedicated single-tenant database pods (separate DB per property) are on the roadmap for Q4 2026 | EU (Frankfurt) | Guest profile data, conversation history, booking records, staff data | GDPR-ready ยท Encryption at rest via platform-managed keys |
Stripe | Payment processing โ tokenised card payments for bookings, room service, and vendor orders | EU & US | Payment method tokens, transaction records (no raw card data stored by Lucy) | PCI-DSS Level 1, SOC 2 Type II |
Google Gemini | Primary LLM โ natural language understanding, conversation generation, reasoning | EU (Google Cloud) | Conversation text, guest queries, context prompts | ISO 27001, SOC 2, GDPR DPA available |
OpenAI | Fallback LLM โ used when primary LLM is unavailable (circuit breaker) | US (with EU API endpoints) | Conversation text, guest queries (fallback only) | SOC 2 Type II, GDPR DPA available |
Fish Audio | Text-to-speech (TTS) โ voice synthesis for Lucy's spoken responses | EU & US | Greeting text and response text for audio synthesis | GDPR DPA available on request |
Twilio | PSTN voice bridge โ optional telephony integration for voice calls | EU & US | Call routing metadata (no call content stored) | SOC 2, ISO 27001, GDPR DPA available |
Google Search (via Gemini) | Live web search โ exchange rates, events, weather, sports scores for guest queries | EU | Search queries for real-time information retrieval | ISO 27001, SOC 2 |
Current deployments use row-level isolation via hotel_id within a shared PostgreSQL instance โ every hotel-scoped query filters by hotel_id, ensuring no guest data crosses hotel boundaries. No data is shared between hotels or used for cross-tenant model training. A dedicated single-tenant architecture (separate database pods per property) is on the roadmap for Q4 2026. Data retention: 90 days post-checkout for conversation data; guest preferences retained per hotel policy. A full data retention schedule and DSAR process is documented in the Data Processing Agreement.
Compliance & Security Architecture
Lucy is one of the few hotel AI platforms that publishes its data exposure model. Every field is classified, every access is logged, every deletion is tracked, and every guest data point is protected by design โ not by policy. Seven compliance pillars, each backed by dedicated backend functions and immutable audit entities. Aligned with GDPR, PCI-DSS, and NIS2 requirements. GDPR tooling is built and operational; PCI-DSS certification and formal WCAG 2.1 AA accessibility audit are in progress.
GDPR Compliance โ UK GDPR & DPA 2018
Lucy is built GDPR-first, not GDPR-adapted. Data minimisation is enforced at the code level โ every memory write passes through enforceDataMinimisation. No voice recordings are stored permanently โ WAV files are deleted after transcription. PII is auto-redacted in all logs via piiAutoDetectionRedaction. The right to erasure is automated via a nightly scheduled scan (automatedGDPRDeletion). Data subject access requests are tracked with a 30-day SLA. Every data access is logged to an immutable audit trail.
- No voice recordings stored permanently โ WAV deleted after transcription
- PII auto-redaction in all logs (piiAutoDetectionRedaction function)
- enforceDataMinimisation โ checks every memory write before it commits
- Automated nightly GDPR deletion scan (automatedGDPRDeletion)
- 30-day SLA tracking for all data subject access requests (handleGDPRDataRequest)
- Consent management per category per guest (ConsentLog entity)
- Immutable audit trail via GDPRAuditLog entity โ every access, every deletion
- GDPR compliance audit: conductGDPRComplianceAudit function โ generates audit-ready report
- Data hosted on Base44 platform infrastructure (PostgreSQL) with encryption at rest
- Encryption at rest via platform-managed keys
PCI-DSS Payment Compliance
All payment processing is handled by Stripe โ Lucy never touches raw card data. PaymentIntent with 3D Secure (3DS) and Strong Customer Authentication (SCA) for PSD2 compliance. Buy Now Pay Later (BNPL) options are available via Stripe's payment methods where supported. As Lucy never stores or transmits raw card data, the platform operates within Stripe's PCI-DSS Level 1 certified environment. Digital tipping via Stripe with full amount passed to staff.
- Stripe PaymentIntent โ Lucy never stores or transmits raw card data
- 3D Secure (3DS) + Strong Customer Authentication (SCA) for PSD2 compliance
- BNPL options available via Stripe's supported payment methods (region-dependent)
- Digital tipping via processDigitalTip function โ Stripe (full amount to staff)
- Operates within Stripe's PCI-DSS Level 1 certified environment
- PaymentTransaction entity tracks every charge, refund, and status
NIS2 Directive Compliance
Lucy complies with the EU NIS2 directive for cybersecurity risk management. Breach notification is automated via notifyNIS2Breach. Security incidents are tracked in a dedicated SecurityIncident entity. The adversarial attack simulator runs red-team penetration tests against the platform. Security posture reports are generated weekly via securityPostureReporter.
- NIS2 breach notification automation (notifyNIS2Breach function)
- SecurityIncident entity โ dedicated incident tracking with severity and status
- Adversarial attack simulator โ red-team penetration testing (adversarialAttackSimulator)
- Weekly security posture reports (securityPostureReporter)
- Rate limiting: per-user throttle (rateLimiterPerUser) + IP-based DDoS protection
- Circuit breaker: automatic LLM failover (Gemini โ OpenAI) on timeout
- Graceful degradation maintains basic service during outages
Voice Identity Verification
When a guest interacts via voice, Lucy is prohibited from accessing or confirming any guest data beyond two fields tied to their currently active reservation. Email is taken from the OAuth token โ never from user input. If no active reservation matches today's date, Lucy declines to confirm any identity information and directs the guest to the front desk. No exceptions, no overrides, no LLM discretion.
- Only guest_name and room_number confirmed via voice โ nothing else
- Email taken from OAuth token, never from user input
- Active reservation filter: check_in โค today โค check_out AND status โ checked_out
- Check-in/out dates, booking reference, room type, dietary data โ all blocked
- Payment status and ID document URLs โ PCI and highest GDPR category, never exposed
- Cross-guest data access architecturally blocked, not just policy-prohibited
Per-Hotel Data Isolation
Every hotel-scoped entity includes a hotel_id field, and every query filters by it โ ensuring no guest data crosses hotel boundaries. Hotel IDs follow a slug format (e.g. 'cafe-royal-london'). Cross-guest data access is architecturally blocked at the query layer, not just policy-prohibited. A dedicated single-tenant architecture (separate database pods per property) is on the roadmap for Q4 2026; current deployments use row-level isolation via hotel_id within a shared PostgreSQL instance.
- hotel_id field on every hotel-scoped entity โ enforced at schema level
- Every query filters by hotel_id โ no cross-hotel data leakage
- HotelConfiguration.hotel_id is the anchor for all tenant isolation
- Cross-guest data access architecturally blocked at query layer
- Dedicated single-tenant database pods (separate DB per property) โ roadmap Q4 2026
Audit-Ready Compliance Trail
Every Lucy action is traceable to a rule, function, or human decision. The AuditLog entity records every state change across the platform. GDPRConsentLog tracks consent per category per guest. GDPRDataRequest tracks DSAR fulfilment with 30-day SLA. GDPRDataDeletion records every deletion event. All audit logs are immutable โ they cannot be modified or deleted by any user, including admins.
- AuditLog entity โ every state change, every function call, every data access
- GDPRConsentLog โ tracks consent per category per guest with timestamps
- GDPRDataRequest โ DSAR fulfilment tracking with 30-day SLA
- GDPRDataDeletion โ records every deletion event with verification
- GDPRAuditLog โ immutable audit trail, cannot be modified or deleted
- Every action traceable to a rule, function, or human decision
Data Security Architecture
Authentication is handled via Base44 OAuth 2.0 with email magic links. Rate limiting is enforced per-user (rateLimiterPerUser) and per-IP (rateLimitingDDoSProtection). A circuit breaker provides automatic LLM failover from Gemini to OpenAI on timeout, with graceful degradation maintaining basic service during outages. Data is hosted on the Base44 platform infrastructure with encryption at rest. Security posture reports are generated weekly, and an adversarial attack simulator runs red-team penetration tests against the platform.
- Base44 OAuth 2.0 authentication + email magic links โ no passwords stored
- Per-user rate limiting (rateLimiterPerUser) + IP-based DDoS protection (rateLimitingDDoSProtection)
- Circuit breaker โ automatic LLM failover (Gemini โ OpenAI) on timeout
- Graceful degradation โ maintains basic service during full outages
- Data hosted on Base44 platform infrastructure with encryption at rest
- Weekly security posture reports (securityPostureReporter)
- Red-team adversarial attack simulator (adversarialAttackSimulator) โ continuous penetration testing
Voice Identity Data Exposure Model
What Lucy can and cannot confirm during a voice session โ published, enforced, auditable. No exceptions, no overrides, no LLM discretion.
| Data Field | Exposed via Voice | Reason |
|---|---|---|
| Guest Name | โ Yes | Required for identity confirmation |
| Room Number | โ Yes | Required for service routing |
| Check-in / Check-out Dates | โ No | Could expose stay duration |
| Booking Reference | โ No | Sensitive reservation identifier |
| Room Type | โ No | Not required for voice verification |
| Dietary Requirements | โ No | GDPR special category โ health data |
| Special Requests | โ No | May contain sensitive personal information |
| Payment Status / Payment Intent ID | โ No | Financial data โ PCI scope |
| ID Document URL | โ No | Biometric / identity data โ highest GDPR category |
| Other Guests' Data | โ No | Cross-guest access architecturally blocked |
If no active reservation matches today's date, Lucy returns verified: false and directs the guest to the front desk. No exceptions, no overrides, no LLM discretion.
Formal Legal Agreements
Downloadable template agreements for enterprise procurement review. Both documents are UK GDPR Article 28 compliant and governed by the laws of England & Wales. Replace bracketed placeholders with your property details and have your legal counsel review before executing.
Data Processing Agreement
GDPR Article 28 compliant. Covers sub-processors, security measures, data retention, breach notification, international transfers, and audit rights.
View DPA
Service Level Agreement
99.5% target uptime, latency targets, incident response (P1โP4), service credit remedy clauses, backup & recovery (RTO/RPO), and termination rights.
View SLA
Both documents are templates โ not legal advice. Have a qualified solicitor review before executing.
UK Regulatory Compliance Suite
Lucy includes a comprehensive suite of 25 UK regulatory compliance modules โ each backed by a dedicated database entity, a dedicated page, and wired into the Hotel Operations Hub navigation. From EICR electrical safety to COSHH chemical registers, from RIDDOR accident reporting to UK GDPR breach management โ your property's compliance is managed, tracked, and audit-ready within a single platform. We are not aware of any competitor that offers an equivalent built-in regulatory compliance suite at this depth.
EICR Register
BS7671 (Electrical Safety)
Fixed wiring electrical installation condition reports. Tracks inspection date, competent person, result (satisfactory/unsatisfactory), certificate reference, C1/C2/C3 defect codes, and next due date (maximum 5 years).
Gas Safety Register
Gas Safety Regs 1998 (CP12)
Gas appliance safety records for boilers, cookers, hobs, water heaters, and fires. Tracks Gas Safe registered engineer, gas safe number, CP12 certificate reference, and next due date.
Fire Safety Log Book
RRFSO 2005
Fire safety checks including alarm tests, emergency lighting, fire drills, extinguisher checks, and fire door checks. Tracks result (pass/fail/partial), location, and next due date.
Fire Risk Assessment
RRFSO 2005
Comprehensive fire risk assessments covering premises, hazards, persons at risk, and control measures. Required under the Regulatory Reform (Fire Safety) Order 2005.
Legionella Temperature Log
HSE ACOP L8
Water temperature monitoring for legionella prevention. Tracks tap type, temperature, in-range status, and corrective actions. Required under HSE Approved Code of Practice L8.
COSHH Register
COSHH Regs 2002
Control of Substances Hazardous to Health. Tracks product name, hazard level, storage location, safety data sheet URL, and first aid measures for every chemical on site.
HACCP Temperature Log
EC 852/2004 (Food Hygiene)
Food safety temperature monitoring for refrigerators, freezers, and hot holding. Tracks temperature, check time, and corrective action when out of range.
Asbestos Register
CAR 2012
Asbestos-containing material register under the Control of Asbestos Regulations 2012. Tracks location, material type, condition, risk level, and management action (manage, encapsulate, remove, monitor).
Accident Book
RIDDOR 2013
Accident recording for guests, staff, contractors, and visitors. Tracks injury type, severity, first aid administered, hospital visit, and RIDDOR reportable status with HSE reference number.
PAT Testing Register
Electricity at Work Regs 1989
Portable Appliance Testing records for all portable electrical equipment. Tracks item name, asset tag, test date, result (pass/fail/visual only), and next due date.
Lift Inspection Log
LOLER 1998 / PUWER 1998
Lift thorough examination and routine service records under Lifting Operations and Lifting Equipment Regulations. Tracks lift identifier, inspection type, result, defects, and next due date.
Pest Control Log
Food Safety Act 1990
Pest control activity and monitoring logs. Tracks pest type, location, action taken, and next visit date. Required for food safety compliance.
Waste Transfer Log
Environmental Protection Act 1990
Waste transfer notes for general, recyclable, food, hazardous, glass, and cardboard waste. Tracks licensed waste carrier, carrier license, disposal site, and consignment note reference.
First Aid Kit Check
Health & Safety (First Aid) Regs 1981
First aid kit inspection records. Tracks kit location, items complete status, missing items, expired items, restocked status, and next check date.
Noise Complaint Register
Environmental Protection Act 1990
Noise complaint tracking for statutory nuisance. Tracks complainant type, noise source, status, and resolution timestamp.
Emergency Command Centre
RRFSO 2005 / NIS2
Emergency protocol management with severity levels, escalation steps, and contact lists. Integrates with NIS2 directive breach notification requirements.
Premises Licence Tracker
Licensing Act 2003
Premises licence, personal licence, DPS, and club premises certificate tracking. Tracks licence number, holder name, issuing authority, conditions, and expiry date.
Health & Safety Policy Register
HSWA 1974
Health and safety policy document register under the Health and Safety at Work Act 1974. Tracks policy title, version, issue date, review due date (must be reviewed annually), and approval status.
Risk Assessment Register
MHSWR 1999
Risk assessment records under the Management of Health and Safety at Work Regulations. Tracks task or area, department, hazards, who is at risk, risk level, and control measures.
Property Walk Audit
HSWA 1974
Property walk audit records for daily, weekly, monthly, and pre-event inspections. Tracks areas checked, defects found, defect count, actions taken, follow-up requirements, and overall rating.
Manual Handling Assessment
MHOR 1992
Manual handling operation risk assessments. Tracks task name, load weight, frequency, risk level, existing control measures, and recommendations.
Lone Worker Monitor
HSWA 1974 / MHSWR
Lone worker check-in system for staff working alone. Tracks staff name, check-in time, location, status, and next check-in time.
DBS Check Tracker
Safeguarding Vulnerable Groups Act 2006
Disclosure and Barring Service criminal record check tracking. Tracks DBS type, check date, result, and renewal due date.
Staff Holiday Manager
Working Time Regs 1998
Staff leave management under the Working Time Regulations. Tracks staff name, department, leave type (annual, sick, unpaid, compassionate, maternity), dates, days off, and approval status.
GDPR Breach Register
UK GDPR / DPA 2018
Personal data breach register as required under UK GDPR Article 33. Tracks breach type, discovery date, severity, containment measures, and notification status.
140+ Operational Tools Across 18 Sections
The UK Regulatory Compliance Suite is part of Lucy's Hotel Operations Hub โ a staff-facing command centre comprising 140+ operational tools across 18 sections. Every tool has a dedicated page, a dedicated entity, and is wired into the hub navigation. This is a code-complete count โ production validation requires the first live deployment.
140+
Operational Tools
18
Sections
25
UK Compliance Modules
0
Gaps Identified (Code Review)
Performance & Reliability
Lucy's architecture was designed against the operational requirements of a 5-star London property โ every architectural decision, latency target, and guest-experience principle on this page was engineered for real-world deployment. Metrics below are from controlled smoke tests (not full-scale load tests โ the n=7 vendor booking test is a functional smoke test, not a stress test). The platform is deployment-ready and undergoing partner onboarding. Full-scale load testing will be conducted during the first pilot deployment.
267
Entity Types (40+ Core)
750+
Backend Functions (200+ Core)
Controlled Load Test Results โ Q1 2026 (Pre-Production)
Vendor booking test (May 14, 2026): 7 bookings across all categories, 100% success, total execution time 6,982ms, all under 3s threshold. This is a functional smoke test (n=7), not a full-scale load test โ it verifies that the vendor booking flow works end-to-end, not that the platform can handle production traffic volume. Latency figures are from this same controlled test. Full-scale load testing (10,000+ concurrent sessions) will be conducted during the first pilot deployment. Production metrics will replace these figures upon first deployment.
Technology Stack
Frontend
React 18 + Vite + Tailwind CSS
PWA, lazy-loaded pages, tanstack-query
Backend
Deno Deploy (Edge Functions)
750+ total functions (200+ core production), independently deployable
Database
Base44 Platform (PostgreSQL)
267 entity types (40+ core), multi-tenancy via hotel_id
Primary LLM
Gemini Flash 3
Via InvokeLLM; sub-100ms latency
Fallback LLM
OpenAI GPT-4o-mini
Circuit breaker on Gemini timeout
Voice TTS
Fish Audio API
Base64 MP3, sentiment-adaptive
Voice STT
Web Speech API + Whisper
Deepgram integration in progress
Live Search
Gemini web grounding
Exchange rates, events, sports scores
Payments
Stripe (PaymentIntent + SCA)
3DS SCA, BNPL via Stripe (region-dependent)
Auth
OAuth 2.0 + magic links
Role: admin | user; public guest pages
Implementation Timeline
From contract to live in 2โ4 weeks. No long IT project. No rip-and-replace.
๐
Week 1
Discovery & Configuration
Hotel profile, brand voice, vendor directory, PMS credentials, room mapping. Lucy HQ panel configured.
โ๏ธ
Week 2
Integration & Testing
PMS sync validated, Stripe payment flows tested, voice pipeline calibrated, staff trained on dashboard.
๐
Week 3
Soft Launch
Lucy live in app for staff testing. Golden Rules enforcement verified. Vendor booking flow tested end-to-end.
โ
Week 4
Full Guest Go-Live
Lucy available to all guests. QR codes in rooms. Post-stay feedback surveys active. Handover to hotel team.
Timeline is a management projection based on a single-property deployment. Multi-property or group-level rollouts may extend timeline.
Become a Founding Hotel
Lucy is built, tested, and deployment-ready. We are now selecting 3โ5 founding hotel partners for the first live deployments. Founding partners receive:
- 50% reduced licence for the first 12 months for full-service hotels (150+ rooms), or 40% for boutique properties (10โ50 rooms) โ applies to either Sponsored or Direct tier
- Direct input into the product roadmap
- Co-marketing opportunities and case study exposure
- White-glove onboarding included (no setup fee)
- Priority access to new features before general release
Founding Partners
3โ5 Slots
First 12 Months
50% Off
To Go-Live
2โ4 Weeks
Setup Fee
ยฃ0
Book a Demo
See Lucy in action at your property. Contact our team to arrange a personalised demo โ not a slide deck, not a recorded video. Lucy handles real guest requests, processes real bookings, speaks 30 languages, and adapts her personality in real time.
Book a DemoSee Lucy at Your Property
Tell us about your property and we'll schedule a personalised demo. No setup fees, no per-room charges.