Skip to content
More support
Contact us
  • There are no suggestions because the search field is empty.

How do I get started with AI and Q-MCP?

This guide gives you ready-to-use prompts, by use case, to connect an AI assistant (Claude, Copilot or ChatGPT) to Q-Data through Q-MCP. Each use case starts with quick prompts you can type as they are, followed by detailed prompts for more demanding deliverables.

What makes this connection possible

Q-MCP is Quinta's connectivity protocol. It allows public AI assistants (ChatGPT, Claude, Gemini, Copilot) to query Q-Data in a standardized and secure way. The AI always responds with certified hotel data: over 3,900 possible data points per property, from pool temperature to parking clearance or pet policies.

With Q-MCP in Quinta, you can:

  • Hotel Search: search for hotels by name, city, country, address, or brand/group.
  • Availability Check: check room availability for specific dates.
  • Hotel Images: get property photos for visual questions (rooms, facilities, etc.).
  • Velma Assistant: ask questions to the AI concierge Velma on behalf of a guest.
  • Guest Data Collection: collect contact information under GDPR consent.
  • Marketing Opt-in: sign up guests to receive promotional emails and exclusive offers.
  • Waitlist: add guests to the waitlist when no rooms are available.

Before you start

1. Connect your AI to the MCP. Follow the connection guide: How to connect Claude to Quinta via MCP. For security reasons, the MCP must be connected once a day.

2. Identify yourself in the first conversation. When connecting for the first time, include this in your first message:

I am [name], my email is [email], phone with country code [+34 XXX XXX XXX]

Some prompts rely on the ask-velma tool, which requires this identification under GDPR consent before answering.

3. Adapt the prompts. The prompts in this guide are written in a generic format. Replace the text in brackets, such as [hotel name], with the actual property or chain data before using them.

 

1. Pre-arrival emails and FAQs

Quick prompt: confirmation email with cancellation policy and check-in times

Draft a confirmation email in [language] for a first-time direct booking,
including the current cancellation policy and check-in times in Q-Data for [hotel name].

 

Detailed prompt: pre-arrival email with pictures

Create a promotional pre-arrival email for [HOTEL NAME], written in English,
addressed to a guest with a confirmed booking.

1. Pull via Quinta MCP:
- hotel profile (search-hotels -> get-hotel-settings): address, neighborhood, phone, email, website, bot languages
- 4-5 real photos via get-media: a "stars" section shot for the hero image, an "accommodation" room, plus exterior/street and restaurant/bar if available.
Use the hosted Q-Storage image URLs directly.
Do not scrape photos from third-party sites.

2. Build an HTML email (table-based, email-client-safe) with:
- hero photo header
- warm welcome with placeholders [Guest name], [check-in date], [check-out date]
- a room photo with caption
- two photos side by side (neighborhood + restaurant/bar or similar) with captions
- a practical info block: address, check-in/out times, parking & access, wifi, contact.
Leave [TBD] for anything the MCP doesn't provide (exact times, parking, wifi).
- a mention of the virtual assistant (Velma) and its active languages, if enabled
- an "upsell" block for amenities (room upgrade, late check-out, in-room treat, private transfer, etc.), tied to the "Special requests" dialog.
No invented prices, just an invitation to ask.
- a closing line inviting special requests

3. Tone: warm, professional, matched to the hotel's category.

4. Save as .html in outputs and present the file.

 

 

Detailed prompt: booking confirmation email with live policies

Draft a booking confirmation email for [HOTEL NAME], for a first-time direct booking guest.

1. Resolve the hotel via Quinta MCP (search-hotels -> get-hotel-settings)
for name, address, contact, and website.

2. For live policy data not exposed by hotel-settings (check-in/out times, cancellation policy, parking):
- Call collect-guest-info with your own name/email/phone under GDPR consent to get a clientId
- Call ask-velma with that clientId, asking:
"What is your cancellation policy, check-in and check-out times, and do you have parking available?"
- Use Velma's real answers instead of placeholders

3. If parking is off-site, find the nearest parking via places_search
and build a simple embedded SVG schematic map (pins for hotel + parking, dashed walking route, walk time).
Inline SVG, not an external image URL,
so it always renders without depending on a third-party map service.

4. Build an HTML email (table-based) with:
- hero photo header
- booking summary block with placeholders [Guest name], [Confirmation number], [Check-in/out date], [Room type], [Rate]
- check-in/out times (real values)
- cancellation policy (real values, with management link)
- parking info + the embedded schematic map
- hotel address and contact
- mention of Velma and available languages

5. Save as .html in outputs and present the file.

2. Front desk and guest relations

 2. Front desk and guest relations 

Quick prompt: answer a guest question at the front desk

Act as the front desk assistant for [hotel name].
A guest asks: "Can we check in at 10:00 AM, and is there an extra cost? Do we need to reserve a table for dinner?".
Answer using real Q-Data.

 

Quick prompt: check a hotel's pet, parking and Wi-Fi policies

Search for [hotel name] in Q-Data
and tell me its configured pet, parking, and Wi-Fi policies.

 

Quick prompt: compare a policy between two hotels of the same group

Compare the early check-in policy between [XXXX] and [YYYY] hotels of the same group
and tell me if there are any inconsistencies between them.

 

3. Newsletters

Quick prompt: subject lines and newsletter body for an offer

Draft three subject line variations and a newsletter body promoting [offer type] for [hotel/group name],
based on the services and conditions configured in Q-Data.

 

Detailed prompt: seasonal newsletter with real content and brand styling

Create a Christmas newsletter for [HOTEL NAME], addressed to guests.

1. Pull via Quinta MCP:
- hotel profile (search-hotels -> get-hotel-settings): address, neighborhood, bot languages
- real content on dining, spa, and activities via get-media (sections: "Restaurant", "Bar", "Spa", or equivalent).
Skip any amenity with no content rather than inventing it.

2. Write the newsletter copy: max 200 words, friendly tone,
highlighting only the amenities confirmed in step 1.

3. Build it as a standalone HTML file with:
- hero photo header with the newsletter title
- 2 real photos side by side (e.g. dining + bar, or dining + activities) with short captions
- a closing line and sign-off from the hotel team

4. Before styling, check the hotel's official website for its graphic charter (colors, typography, visual style) and apply it.
If the site isn't accessible, search press coverage for design cues instead
and flag it as a best-effort match, not an exact extraction.

5. Save as .html in outputs and present the file.

 

4. Fact sheets and RFP responses

Quick prompt: two-page fact sheet for travel agencies

Generate a two-page fact sheet for [hotel name] targeting travel agencies:
room types, dining, wellness, transportation, and contact details from Q-Data.
Deliver it in PDF format.

 

Quick prompt: respond to an RFP

Draft a response to this RFP request for an event with [number] people from [date] to [date]:
[paste RFP content].
Use room capacity, catering, and guest room data for [hotel name] from Q-Data.

 

Quick prompt: check data completeness before sending a proposal

Before sending the proposal, verify that key data fields for [hotel name]
are complete and updated in Q-Data.
Highlight any empty or outdated fields that could weaken the proposal.

 

Detailed prompt: two-page PDF fact sheet built only from Q-Data

VARIABLES
- Hotel: [HOTEL NAME]
- Audience: [travel agencies / MICE and groups / corporate / OTA partners]
- Document language: [English / French]

MISSION
Produce a two-page PDF fact sheet for this hotel, written for the audience above.
Every fact must come from Q-Data through the Quinta MCP PROD connector,
using only three tools: search-hotels, get-media and ask-velma.
No external source, no website content, nothing invented.

This property may not have a restaurant, a bar, a spa, a gym, parking or meeting space.
Nothing in this brief is a section you owe me.
Build the document from what exists, and only from what exists.

STEP 1 - IDENTIFY THE PROPERTY
Call search-hotels with the hotel name, stripping generic words such as "hotel" or "resort".
Report back to me in chat: property_id, group_id, star rating, address, website URL and the logo photo_url.
Use that property_id in every later call.
If several properties match, list them and ask me which one before going further.

STEP 2 - ESTABLISH WHICH SERVICES ACTUALLY EXIST
Before writing anything, build a service inventory.
Ask Velma plainly whether the property has each of the following, one question per item:
restaurant, bar, room service, breakfast, spa, gym or fitness room, swimming pool,
meeting or event space, parking, airport transfers, laundry.

Sort every service into exactly one of three buckets
and show me the three buckets in chat before you build the PDF:
- EXISTS AND DOCUMENTED: the service exists and Q-Data has usable content for it. It gets a section in the PDF.
- EXISTS BUT EMPTY: the service exists but Q-Data has no usable content. It gets at most a one-line mention, and it goes in your data-gap report to me.
- DOES NOT EXIST: the property does not have it. It gets no section, no heading and no placeholder.

An empty get-media section is not proof that a service is absent.
It may simply have no photos loaded.
Confirm absence with Velma before concluding it,
and never write "no spa" on the strength of an empty media call alone.

State an absence in the PDF only when the audience would otherwise assume the service is there:
no on-site restaurant in a city hotel, no parking in a destination people drive to, no lift in a historic building.
Keep it to one factual line, with the nearby alternative if Velma provides one.
Otherwise say nothing at all.
A fact sheet that lists what a hotel lacks reads as an apology.

STEP 3 - PULL THE MEDIA CONTENT
Call get-media on the "Accommodation" section.
Use it for the room categories and for the real descriptions carried in the photo captions.
Do not paraphrase a caption into something more marketing-sounding than it is.

Then call get-media only on the sections your step 2 inventory says exist,
typically "Restaurant", "Bar", "Spa" or their local equivalents.
Do not call sections speculatively and do not treat the list above as a checklist to fill.

Collect every URL returned by get-media: room galleries, detail pages, menus, booking links.
They become the clickable links in the PDF.

STEP 4 - ASK VELMA FOR THE REST
Everything not covered by get-media comes from ask-velma.
If the tool requires guest identification first, present the privacy notice verbatim,
wait for my explicit agreement, then collect name, email and phone before asking anything.

Ask one targeted question per data point rather than one broad question.

Always ask:
- room types with the number of rooms, size and maximum occupancy per category
- check-in and check-out times, early check-in and late check-out conditions
- city tax, service fees, VAT treatment
- rate plans available
- cancellation policy per rate plan
- reservation and front desk contact details
- languages spoken and guest service languages
- accessibility for guests with reduced mobility
- distances and travel times to airports and stations

Ask only if step 2 confirmed the service exists:
restaurant cuisine, seating and service hours; bar hours and capacity;
spa facilities, hours and signature treatments; meeting room capacities;
parking type and price; transfer pricing.

Ask only if the audience is trade:
group booking threshold, travel agent commission, GDS codes, OTA property IDs, dedicated trade contacts.

Quote the answers faithfully.
If Velma answers that information is not available, record it as a gap and leave it out of the PDF.
Do not substitute your own knowledge.

STEP 5 - DATA QUALITY CHECK BEFORE WRITING
Report to me in chat, not in the PDF:
- any mismatch between the declared total room count and the sum of the category counts
- any obviously templated or placeholder content still sitting in Q-Data, for example bracketed strings such as "[Insert phone number]", or amenities that are implausible for this property. Exclude those blocks from the PDF
- any mapping that cannot be resolved, for example room views stored against UUIDs that do not resolve to readable room names
- every service in the EXISTS BUT EMPTY bucket, since that is a configuration gap the hotel should close and it is also degrading Velma's answers to guests
- every question where Velma had no answer
Never invent to fill a gap.

STEP 6 - VISUAL IDENTITY
Take the hotel's official website URL from step 1, download its stylesheet and read:
the primary colour, the light or beige background tones, the body text colour,
and the font families declared in the @font-face rules.
This is the only step that touches anything outside Q-Data,
and it is for styling only, never for content.

Apply those colours to the header band, section headings, table header fills,
alternating row shading, rules, links and the footer bar.
Use the official logo from the photo_url returned in step 1,
trimmed and composited on the header background.

If the site fonts are commercial, for example Gotham, Brandon or a licensed didone, do not embed them.
A web font licence does not cover embedding in a distributed PDF.
Use a close free substitute, tell me which one and why,
and keep the generation script parameterised so the real fonts can be dropped in later.

STEP 7 - PDF STRUCTURE
A4 portrait, exactly two pages, no more.

Treat the document as modular blocks, not a fixed template.
Render a block only if it has content.
Never print a heading with nothing under it,
and never write "not available" under a heading to keep the layout symmetrical.

Fixed blocks, always present:
- header band with logo and hotel name
- quick facts: address, districts, category, inventory, check-in, check-out, front desk, languages
- room table: type, category, number of rooms, size, maximum occupancy, bedding, links column
- rates, taxes and policies
- location and access
- contacts
- source note

Conditional blocks, included only if step 2 put them in EXISTS AND DOCUMENTED:
dining, bar, wellness, meeting and event space, parking and transfers,
sustainability credentials, guest ratings.

Fill the two pages by expanding the blocks that do exist rather than by adding empty ones.
Room detail, accessibility, neighbourhood and access, and rate conditions
all take more depth on a property with few services.
If after that there is genuinely not enough content for two pages, stop and tell me.
Produce a well-composed single page instead of a padded second one, and say why.

Weight the content for the audience.
Trade: commission, GDS codes, cancellation terms.
MICE: capacities, meeting space, commercial contacts.
Corporate: rates, cancellation flexibility, transfers and connectivity.

STEP 8 - SOURCE NOTE
Close the last page with a small-type source note stating:
sourced entirely from Q-Data through Quinta MCP PROD, the generation date,
the property_id and group_id, the tools used,
that no external source and no editorial invention were involved,
and that rates and policies should be reconfirmed with the hotel before contracting.

STEP 9 - DELIVERY
Save the file as .pdf in outputs and present it.
Before handing it over, verify: the page count matches what you told me,
the links are clickable, no text overflows or collides with the footer,
and no section heading sits above empty space.
Then summarise for me in a few lines: which services exist and were documented,
which exist but had no content, which do not exist,
and the data anomalies the hotel should fix.

 

5. Group proposals and MICE

Quick prompt: find hotels matching event criteria

Search which hotels have a conference room for more than [number] people, a pool, and parking.
Generate a comparison table.

 

Quick prompt: group proposal for an event

Using Q-Data for [hotel name], draft a proposal for a group of [number] people
([event type: wedding, corporate retreat, conference]),
including room types, menu options, and available photos.

 

Quick prompt: multi-hotel group proposal with transfer cost

Generate a group proposal for [number] adults,
with a potential split between [XXXX] and [YYYY] hotels if one lacks full availability,
and include the estimated airport transfer cost from [XXXX].

 

6. Landing pages

Quick prompt: SEO and GEO landing page for a hotel

Create a landing page for [hotel name], optimized for SEO and GEO.
Include [number] photos, a short description of [number] characters,
a list of featured amenities, and the dynamic Q-SEO FAQ section.

 

Quick prompt: group landing page combining two hotels

Generate a group landing page for [event type],
combining data from [XXXX] and [YYYY] hotels.

 

7. Audit and data consistency

Quick prompt: compare Q-Data, website and OTA data

Conduct a comparative study of [hotel name],
comparing its Q-Data, website [URL], and [OTA, e.g., Booking.com] [URL].
Indicate whether the data is consistent
and provide a prioritized action list to correct discrepancies,
using Q-Data as the canonical reference.
Output in HTML.

 

8. Upselling and revenue

Quick prompt: three upselling options for a reservation

Based on the services configured in Q-Data for hotel [ID],
suggest three upselling options (room upgrade, spa, F&B)
for a guest with a [room type] reservation for [number] nights.

 

9. Onboarding and internal training

Quick prompt: train a new front desk staff member

Act as a trainer for a new front desk staff member at [hotel name].
Explain the hotel's check-in/check-out, pet, and cancellation policies using real Q-Data.

 

Quick prompt: internal FAQ guide for guest relations

Generate an internal FAQ guide (not for guests) regarding hotel services for [ID],
aimed at onboarding guest relations team members.

 

How to get the most out of it

These prompts can be combined into one when the task allows. For example, you can request a hotel search, review a specific policy, and draft a guest response within the same message. The more context you provide, such as hotel ID, language, desired tone and output format, the more accurate the output will be.

The detailed prompts follow a common pattern you can reuse for your own use cases: identify the property first, pull real content only from Q-Data, leave [TBD] or report a gap rather than inventing, and specify the output format and where to save it.