AI FOR OPERA
D3x extends automation from the front desk to every guest request: it understands a guest on any channel and completes the action in your systems — natively on Oracle OPERA Cloud via OHIP, not another chatbot that answers and hands off.
Live across 60+ European hotel groups · 25+ integrations · 4.58 AI CSAT
01 — What D3x does
Most guest AI stops at the answer — it informs, then routes the guest to a link or your team. D3x goes the last mile: it does the thing, in your stack, and tells the guest it's done.
01
Reads guest intent on any channel, in every language you operate — “late checkout?”, “change the name on my invoice”, “the AC isn't working.”
02
Decides the right action against your rules — rates, availability, policy, brand voice — and checks what's possible right now.
03
Completes it in OPERA and the operational stack — reservation, folio, housekeeping, profile — then confirms back to the guest.
Why this fits enterprise Opera estates
Multi-property groups run on automation and lean front desks. Resolving even a slice of guest requests end-to-end removes thousands of staff touches a week — the same efficiency logic, now applied to guest service across every property on OPERA Cloud.
The cost of not answering
On a direct-booking model, guests who can't get through don't wait — they book on an OTA, hand away the relationship and pay commission. The compounding cost is larger than it looks.
20–30%
Peak-time reservation calls go unanswered (industry benchmark)
60%
Of contacts resolvable with zero human touch
< 2 min
Time before a digital enquiry abandons (budget / midscale)
High-intent callers reach voicemail on busy evenings and weekends. A call lost is often an OTA booking gained — commission paid and the guest relationship handed away.
Web chat, WhatsApp and email enquiries not answered quickly become abandoned bookings. Budget guests compare and move on within minutes.
Conversational and Gen-AI trials prove appetite but a PoC answers — it doesn't resolve in OPERA. Booking, billing, modifying across every channel is the missing step.
Guest-contact volume grows faster than front-desk headcount. A lean, fast-growing estate can't absorb that load linearly without either cost or service slipping.
D3x measures real figures — calls answered, messages resolved, containment rate — per property, from the first week live.
02 — Across every channel
The same intelligence works web chat, WhatsApp, the phone and the inbox — and carries context between them. A few everyday requests, resolved end-to-end:
“Table for four at the restaurant tonight, around 8?”
Booked — 8:15pm is the closest slot, table for four under your name. I've noted it on your stay and you'll get a reminder at 7. Anything dietary I should flag?
→ restaurant.availability ✓ → reservation.create ✓ → reservation.note ✓ → PMS
“I think I left my laptop charger in room 312.”
Found it — housekeeping has your charger from 312. I've logged it to your profile and can post it to your home address today, or hold it at reception. Which would you prefer?
→ housekeeping.lostfound.match(312) ✓ → profile.attach ✓ → shipping.quote ✓
“Can you reissue my invoice to my company, Lefevre SARL, instead of my name?”
Of course — I've updated the billing profile on your booking to Lefevre SARL and reissued the invoice. The new PDF is on its way to your email, and the personal copy has been voided.
→ pms.reservation.find ✓ → profile.billing.update(company) ✓ → invoice.reissue ✓ → email.send
“Can you arrange a car to the airport at 6am on Friday?”
All set — a private transfer is booked for 6:00am Friday from the lobby, billed to your room. Your driver's details will arrive by message the night before.
→ transfer.book ✓ → reservation.note ✓ → folio.charge ✓ → comms.confirm
One agent, one memory — every request resolved in your systems and confirmed back to the guest.
Agentic email
D3x's email agent reads the full thread — with the reservation and folio pulled from OPERA — then drafts, books, modifies or routes, with a one-click approve on anything high-stakes.
Booking changes, invoice requests, confirmations, group enquiries and partner emails get read, understood and actioned — not just triaged and forwarded to a person.
It reads the entire thread and pulls the reservation and folio from OPERA before acting — no copy-pasting between systems, no re-reading a chain to find what the guest actually asked for.
Drafts and executes — modify a booking, reissue an invoice, route a group lead — and holds anything sensitive for one-click human approval before sending.
Support and back-office people spend the day on exceptions, escalations and relationships — not hundreds of near-identical emails.
Currently in beta, running in co-pilot mode where staff approve before anything sends.
Enterprise groups on OPERA + D3x
From luxury resorts to city aparthotels, enterprise operators run D3x as the guest-facing layer on top of Oracle OPERA Cloud — one connection, the whole portfolio.
L&R Group
Part of L&R Group — landmark luxury hotels and resorts where D3x carries VIP context, spa and dining across every channel without ever breaking brand.
Ireland's largest hotel operator, running Clayton and Maldron across Ireland, the UK and mainland Europe — guest requests resolved end-to-end on OPERA Cloud.
Staycity and Wilde aparthotels across Europe — long-stay check-in, access codes and extensions handled automatically, freeing lean front desks at scale.
03 — The guest concierge
The same orchestration, dressed in your own look — a concierge your guests recognise as yours, that resolves requests and quietly grows revenue.
Illustrative · your-branded concierge
Guests meet your property — your name, your tone — not a generic bot bolted on. White-label the experience; D3x runs the orchestration underneath.
In the same breath as a helpful reply, it surfaces direct-booking perks and add-ons — direct-booking lift and ancillaries, captured in the conversation.
The guest chooses how to talk; D3x carries the context across all of them — including straight into the voice call.
Behind the concierge, D3x books, bills, updates and dispatches in your systems. It resolves — it doesn't just reply.
04 — Voice AI · on your property lines
Callers on your property line are high-intent — closer to booking than a website visitor. Yet a large share of inbound calls go unanswered at peak times, nights and weekends. D3x Voice answers every inbound call — resolving the request, capturing the booking, or handing off to your team with full context.
20–30%
Peak-time reservation calls go unanswered
€165K+
Est. direct revenue lost per property, per year
€1,450
Average voice-channel booking value
Industry estimates — Revinate 2024 hospitality voice benchmark and published analyses, converted to EUR; figures vary by property size, season and staffing.
Legacy PBX / IVR
It routes the call. It doesn't resolve it.
D3x Voice
Same Skills Engine as messaging — no separate voice setup.
D3x Voice sits on top of your existing telephony and reuses the D3x knowledge base you already run — no rip-and-replace, no separate voice setup. It can go live on a property line overnight, then take over IVR routing as you roll out across the estate.
In production: 10,000+ hours of guest voice handled, scoring 4.58 AI CSAT against a 4.62 human benchmark across 20,000 surveys. For lean front desks across a multi-property estate, every line is answered — peak, nights and weekends — so no high-intent call leaks to an OTA.
05 — Native to your stack
D3x is built natively on the Oracle Hospitality Integration Platform — reading and writing reservations, folios, profiles and availability, and coordinating housekeeping and CRM. It sits on top of your stack through 25+ live integrations; nothing is ripped out.
Guest channels
D3x — orchestration
Understand intent · Apply your rules · Act & confirm
OHIP · Oracle Hospitality Integration Platform
Your systems of record
HousekeepingCRMMobile appPaymentsAcross hundreds of properties, D3x deploys on one connection — not hundreds of bespoke ones. New deployments are live in about three weeks.
The difference that matters: D3x doesn't only look things up, it commits the change to the system of record — the prerequisite for resolving, not just answering.
Standardise on OPERA Cloud or run a mixed estate — D3x is built OPERA-native and sits on top either way.
06 — Chatbots vs. orchestration
Typical hotel chatbots are built to inform and route. D3x is built to resolve and execute. Both talk to guests well; only one acts on your systems.
| Capability | Typical chatbot | D3x |
|---|---|---|
Answer guest questions FAQ, info, multilingual, 24/7 | Strong | Strong |
Drive direct bookings Guide the guest into the booking flow | Strong | Native |
Resolve a request end-to-end Complete the action, not just describe it | Hands off | Native |
Write back to OPERA Modify reservations, folios, invoices | Rare / read-only | Deep · OHIP |
Phone & voice Resolve on the call vs. route / callback | Route + book | Resolve |
Omnichannel memory One context across web / WhatsApp / phone | Per-channel | Unified |
Housekeeping & operations Act on room / service status | Out of scope | Native |
CRM & guest-profile actions Update profiles, preferences, billing | Limited | Native |
Self-healing knowledge Stay correct without manual upkeep | Manual database | Self-healing |
OHIP-native orchestration Coordinate across the OPERA ecosystem | No | Built for it |
60%
Containment · zero-touch
120×
Faster response · 15m → 7s
4.58
AI CSAT
+95%
Volume · no new hires
Built for enterprise IT
D3x is built for multi-property groups and the teams that have to sign off on them — security, control and reporting that fit how enterprise hospitality IT actually operates.
GDPR-compliant by design, SOC 2 Type II in progress, EU data residency, encryption in transit and at rest. A Trust & Security pack ready for procurement.
Define exactly what agents may do autonomously, where they must hand off to a human, and which actions need approval — brand-voice and policy guardrails enforced on every reply.
Role-based access control, full audit logs and SSO support, with SCIM provisioning on the roadmap — so the right people see the right properties, and every action is traceable.
A clean hierarchy of brand-level and property-level accounts with scoped roles and permissions — manage the portfolio centrally while each property keeps local control.
Portfolio-wide reporting out of the box: containment, response time, CSAT, revenue captured and per-property performance — exportable for your BI stack.
REST APIs, webhook-driven event flows and an open platform (MCP) — fully managed, with daily LLM upgrades and new integrations shipped to your teams at no extra effort.
One OHIP connection for the whole estate, an observability layer with health checks and partner SLAs, and a fully managed service — no engineering team required to keep it running.
07 — Knowledge that maintains itself
Every property has its own parking, breakfast and check-in rules — in a dozen languages. The question for any guest AI at scale is who keeps all of that true.
Curated & re-read
Staff load answers per property; the AI re-reads its sources on a schedule.
When reality changes — a closed car park, a new breakfast time — it's only as right as the last manual edit.
Gaps surface as wrong answers to guests, caught reactively and fixed by hand.
Multiply the upkeep by hundreds of properties in many languages and it becomes a standing operational cost.
Accuracy depends on how diligently every property keeps a database up to date.
Reconciles to source of truth
Captures every escalation, correction and retrieval gap as structured signal.
Diagnoses the root cause — missing policy vs. stale source vs. PMS conflict.
Proposes the smallest safe knowledge update, drafted automatically.
An operator approves — then production updates. Knowledge converges, not drifts.
Accuracy becomes a property of the architecture, not a daily chore at every hotel.
This is the quiet differentiator at chain scale: the AI stays correct on its own.
08 — AI-optimised upsell
Before each arrival, D3x works out the right add-on for each guest — early check-in, late checkout, breakfast, parking — and sends it on WhatsApp, read about 98% of the time, against roughly 20% for email. The guest taps once; D3x books and bills it.
€5–12k
Per hotel · per year
€5–10M
Across ~900 hotels · per year
~5×
Reach of WhatsApp vs email (98% vs ~20% read)
How the estimate is built — per hotel, conservative → moderate
Illustrative and deliberately conservative, built on published industry benchmarks (Oaky pre-arrival upsell, Revinate 2024, WhatsApp Business 2025). Actual results vary with reachability, opt-in, offer mix and season — D3x reports real per-property conversion once live.
09 — Let's talk
D3x is the layer that turns guest conversations into completed jobs — across every channel, in your systems, staying accurate on its own. It extends the automation you already run rather than replacing anything.
24/7
Upsell · every request answered, every add-on surfaced
4.58
CSAT · live AI guest-satisfaction score
60%
Containment · resolved with zero human touch
01 · Connect
Reuse the knowledge you already have. Nothing is ripped out — D3x sits on top of your existing stack.
02 · Go live
Days, not quarters. Real guest traffic from day one — calls answered, messages resolved, add-ons captured.
03 · Measure
Calls answered, messages resolved, containment rate, add-on revenue — per property. Once the numbers are in, roll the estate on one integration.
Bring a couple of everyday requests — a late checkout, an invoice change, an “AC not working” — and we'll map them live, end to end, against your OPERA stack. No slideware: your scenarios, your systems.
Jason · Founder & CEO, D3x · jason@d3x.ai · d3x.ai