How to set up Q-MCP?

What is Q-MCP?

Q-MCP lets an AI assistant - like ChatGPT, Claude, or Copilot - safely use your hotel's data to answer guest questions, check room availability, capture leads, and show your property photos. Your data stays under Quinta's control at all times: the assistant gets the answers it needs without ever receiving your raw content.

In short, it makes your hotel's information usable inside AI assistants and apps, without ever handing over the underlying data.

What can you do with Q-MCP?

Here are the most common ways our clients use it:

Guest Experience:

Grounded pre-arrival and info emails: Connect your AI assistant to Q-MCP and pull a hotel's narrative summaries (arrival & departure, breakfast, Wi-Fi, booking terms) straight from Q-Data to draft accurate pre-arrival or FAQ emails.
Accurate complaint responses : Before drafting a reply to a guest complaint, have the assistant retrieve the actual conversation history for that guest on Velma.
Internal concierge assistant: Give front desk and guest relations staff a Claude or Copilot assistant connected to Q-MCP. They ask it "Is the spa open on Sundays?" or "What's our late check-out policy?" and get an answer sourced directly from the hotel's own configuration.

Sales and Marketing:

Creating marketing and sales content: your team uses an AI assistant connected to Q-MCP to draft newsletters, factsheets, and RFP responses, or to build a landing page, all based on accurate, up-to-date hotel data.
Portfolio search for group and MICE proposals: When a client brief calls for specific features (conference rooms, pool, spa, parking), search across the whole group to shortlist matching properties in seconds, instead of checking each hotel individually.
Upsell opportunities from real conversations: Search recent guest messages for recurring requests (late check-out, extra bed, spa reservation) to spot demand patterns worth turning into a standing upsell offer.
Photos inside a partner app: a partner shows your property's photos (rooms, spa, pool, restaurant) inside their own app, with no guest data shared.

Operations:

Cross-property consistency checks: For chains, compare dialog configuration across properties in a group to spot where one hotel's operational data has fallen out of sync with the rest.
Configuration completeness audits: Run a check across a property's operational dialogs (room equipment, services, opening hours) to flag which fields are still empty or on a generic default, so your teams know exactly what to fill in.

How do I get access?

Everything happens in your console: https://console.quinta.im/channel/mcp. To set up access, you need to open it with an Owner or Manager role.

There are two ways to connect Q-MCP, depending on who - or what - will be using the tools.

Option A: API key (for applications and automated tools)

Use this when a website chatbot, mobile app, or third-party tool will connect to Q-MCP automatically.

Click Create API key
Select which building blocks (tools) the application should be allowed to use — only enable what the use case requires
Set an expiry date — never create a key with no expiry
Share the key securely with your developer or technology partner

Security note: treat your API key like a password. If a key is compromised, revoke it immediately and generate a new one. Rotate keys regularly.

Option B: MCP user account (for your team's AI assistants)

Use this when a person on your team will connect Q-MCP directly to their Claude, Copilot, or ChatGPT account for day-to-day use.

Click Create MCP user
Select which building blocks the user should be able to access
The user connects their AI assistant directly from the console — no API key to manage

Once either type of access is created, Quinta runs a pre-launch check before the connection reaches any guest (see the validation section below).

For developers and IT teams:
What data can I retrieve?

Q-MCP exposes eight tools, grouped into four functions. Enable only what your use case requires.

Function	Tools
AI assistant	ask-velma
Guest identity & booking pipeline	collect-guest-info, get-availability, marketing-opt-in, waitlist
Multiproperty routing	search-hotels
Media	get-media

Here is what each building block does:

Building block	What it does
Ask Velma	Passes a guest's message to Velma and returns a ready-made reply, including visual carousels for rooms, bars, and spas. Velma answers the specific question from your hotel's information - it never hands over your full knowledge base, so your content can't be copied in bulk or used to train someone else's AI. It only works once a guest has been registered with consent (it must be paired with guest sign-in below), otherwise it won't respond.
Guest sign-in & consent	Registers the guest with their privacy consent and creates the guest ID that every guest-facing and booking action needs. It's the starting point for anything guest-facing, and the required partner to Ask Velma.
Room availability	Checks whether rooms are free for the guest's dates. When combined with guest sign-in, each availability check creates a tracked lead in Q-Sales. When the guest then books, Q-MCP marks that lead as confirmed - so you can see real, measurable revenue coming from the assistant.
Marketing consent	Records the guest's agreement to receive marketing messages during the same conversation. Optional, and needs guest sign-in first.
Waiting list	When no rooms are available, adds the guest to your waiting list as a lead. Like the chat widget, it only works with booking engines connected by API (a direct software link). Optional.
Multi-property routing	Helps the assistant figure out which of your hotels the guest means and send them to the right one. Required for any setup covering more than one property - without it, the guest-facing and booking blocks won't work. Single-hotel setups don't need it.
Property photos	Returns your property's stored images from Q-Storage - room types, spa, pool, restaurant, and more - main photo first. It lets the assistant show, not just describe, when a guest asks to see a room. It carries no guest data and is safe to share.

Which building blocks must be used together?

Some blocks only work in combination. Turn them on together, or the experience can behave unpredictably for your guests.

Single-hotel guest assistant (minimum): Ask Velma + guest sign-in.
Optional: property photos.
Multi-property guest assistant: Ask Velma + guest sign-in + multi-property routing. Optional: property photos.
Guest assistant with tracked bookings: add room availability.
Optional: marketing consent, waiting list.
Photos only: property photos.

What breaks if configured incorrectly?

Ask Velma without guest sign-in: Will not respond
A multi-property setup without multi-property routing: Won't work.
Room availability without guest sign-in: Works, but no lead is created in Q-Sales.
Waiting list without an API-connected booking engine: Works, but no waiting-list lead is created.
Set up correctly: Works as expected.

Which building blocks should I turn on for my use case?

Website or app guest assistant (single hotel): Ask Velma, guest sign-in, property photos, room availability, marketing consent (add waiting list if your booking engine is API-connected).
Chain or group guest assistant: all of the above + multi-property routing.
Photos inside a partner app: property photos only.
Q&A pilot, no booking: Ask Velma + guest sign-in.
Content creation (newsletters, factsheets, RFPs, wedding/group landing pages): property photos plus the data blocks relevant to your content - no guest-facing blocks needed.

Example: Marketing and Sales content

A sales team uses Claude with Q-MCP to generate factsheets, RFP responses, and group landing pages from live hotel data. A sales manager connects Q-MCP and asks: “Draft a two-page factsheet for [Hotel Name], including room types, capacity, spa services, F&B options, with photos.” Claude calls Velma for property data and get-media for images, then returns a complete, accurate draft—no manual lookup needed.

Tools to enable: get-media + ask-velma + collect-guest-info

How is my setup checked before going live?

Before your setup reaches a guest, we test the real use case end to end. Here's what we check:

Access & permissions: the right access has been created, covering only the blocks you intended to enable.
Data check: what each block returns is clean, suitable for your use case, safe to share, and protects your interests.
A real run-through: we try your actual scenario, not a quick test. For a guest assistant, that means a full guest sign-in followed by a real conversation.
Your confirmation: for setups tied to your own account that we can't test directly, we ask you to confirm everything works.

To test a connection independently, you can use MCP Jam: https://www.mcpjam.com (log in with a Google account). It's the quickest way to confirm a setup is responding as expected.

How does Q-MCP protect my hotel and my guests' data?

Velma gives answers, not the data behind them: your hotel's information is never handed over to the outside AI in bulk, which protects it from being copied or used to train another company's AI.
Guest-facing conversations only start after the guest has given explicit consent through guest sign-in.
The photo block carries no guest data. The guest sign-in and booking blocks do - so we only turn them on where your use case needs them, keep them tightly limited, and never put them behind an access code that never expires.