Working MCP demonstration

Live, focused and read-only

Ask ChatGPT. Find the right room.

QuantaDuck Room Finder demonstrates how ChatGPT can securely query organisational systems through a purpose-built MCP server.

Add to ChatGPT See example questions
Live demonstrationRead-onlyFictional college data

End-to-end proof of concept

A small integration that really works.

The demonstration connects natural language in ChatGPT to a narrow, validated search tool and genuine availability calculations—all without exposing a booking action.

01

Natural-language requests

Users ask questions normally instead of completing a complex search form.

02

Structured MCP tool

ChatGPT calls find_available_rooms with validated dates, times, capacity and filters.

03

Availability logic

The service checks room characteristics and recurring occupied periods before returning a match.

04

Safe, read-only results

The tool can search and explain results, but it cannot create or change bookings.

Live service checkThe page checks the public health endpoint without blocking the experience.
Checking MCP service…

Try these questions

Start with ordinary language.

Copy a question, invoke QuantaDuck Room Finder in a ChatGPT Work conversation, and see the structured room search in action.

QUESTION 01

Find a computer room for 20 learners on 25 August 2026 from 10:00 to 12:00.

QUESTION 02

Find an accessible classroom for 30 people at Main Campus on 26 August 2026 from 13:00 to 17:00.

QUESTION 03

Find a science lab with a projector for 18 learners between 09:00 and 11:00 on 27 August 2026.

QUESTION 04

Are any rooms available for 200 people on 27 August 2026 from 09:00 to 11:00?

SAFE BOUNDARY

Book the first available room.

The plugin will refuse: this proof of concept is deliberately read-only.

How it works

One request. Four controlled stages.

ChatGPT handles the conversation. The MCP server exposes one narrow capability with a defined input and output contract.

Ask

The user describes the room they need in ChatGPT.

Understand

ChatGPT converts the request into date, time, capacity, campus, room type and features.

Search

ChatGPT calls the QuantaDuck MCP endpoint, which runs the availability logic.

Answer

Matching rooms return to ChatGPT and are explained conversationally.

HostingCloudflare Workers
ProtocolModel Context Protocol
TransportStreamable HTTP
Toolfind_available_rooms
AccessNo authentication · fictional data
ModeRead-only · no API key required

How the connection works

A demonstration today. A governed integration tomorrow.

The current service proves the interaction pattern. It does not connect to a college system, process personal data or imply that production integration is simply a URL change.

Demonstration today

Deliberately self-contained

A safe public example designed to show the complete MCP journey.

  • 12 fictional college rooms
  • Fictional recurring occupied periods
  • Data compiled into the application
  • No college-system access
  • No personal data or bookings
  • No OpenAI API call or API key
Production integration

Connected to approved systems

The same focused tool pattern could work with authenticated organisational APIs and services.

  • Room and timetable systems
  • Student or curriculum systems
  • Booking platforms and CRM systems
  • Reporting databases
  • Internal business applications
  • Narrow tool access—not unrestricted system access

Add it to ChatGPT

Connect the working MCP server.

Developer mode and plugin availability can depend on your ChatGPT plan and workspace administrator settings.

  1. Open ChatGPT on the web.Use your usual ChatGPT account and workspace.
  2. Open Settings → Security and login.Find the developer settings for your account.
  3. Enable Developer mode.Availability can depend on your plan and workspace settings.
  4. Open ChatGPT Plugins.Go to the plugin management area.
  5. Select the plus button.Create a developer-mode plugin.
  6. Enter the MCP server URL.Use the endpoint shown alongside these steps.
  7. Name it QuantaDuck Room Finder.Complete the connection details.
  8. Open your personal Plugins directory.Open the plugin and install it if prompted.
  9. Start a new Work conversation.Return to ChatGPT and switch from Chat to Work.
  10. Type @ and choose QuantaDuck Room Finder.Then try one of the example questions above.

What this proves

Room finding is only the demonstration.

The reusable idea is allowing ChatGPT to perform a narrow, controlled task using an organisation’s existing data or services. Each production integration would be designed around the customer’s systems, permissions and governance requirements.

01Check learner timetables
02Find available staff or resources
03Look up course information
04Produce controlled management reports
05Check booking availability
06Search internal policies and procedures
07Start approved business workflows

For colleges and businesses

Could ChatGPT work with your systems?

QuantaDuck builds focused integrations that connect conversational AI to the tools and data organisations already use.

Talk to QuantaDuck