empower

• 

• 

• 

Inspiration

In North America alone, the hospitality industry has been dealing with serious labor shortages since COVID. According to the American Hotel & Lodging Association, around 65% of hotels report staffing shortages, and many say they’re unable to fill open positions. Even in Canada, hotels have struggled to fully rebuild their teams post-pandemic.

At the same time, guests expect instant responses. If someone messages through the website, Booking.com, or Instagram, they don’t want to wait 6 hours. They want help now. Empower/Mage: Hotel Guest Communication Interface

Mage is a mobile-first web application designed to streamline hotel guest requests, automate routine inquiries, and facilitate seamless handoffs to human staff. Built with a Next.js frontend and a FastAPI backend, it provides a highly deterministic, state-driven chat interface.

What Problem It Solves

Hotel front desks frequently experience bottlenecks due to high volumes of routine inquiries (e.g., WiFi passwords, check-out times, amenity hours) and minor service requests (e.g., extra towels, room service).

Mage solves this by acting as the first layer of guest interaction:

1. Deflection of Routine Queries: Instantly answers common questions using a deterministic intent layer and a scoped knowledge base.
2. Automated Ticketing: Parses service requests (maintenance, housekeeping) and automatically generates tickets in the hotel's property management system.
3. Frictionless Escalation: Transitions guests to a human front-desk agent when requests fall outside the automated scope or when specifically requested.

Architecture

The system is split between a strictly typed React frontend and a Python backend handling routing, transcription, and database operations.

Frontend Framework: Next.js / React

State Management: Zustand (Application State) & React Query (Server State)

Styling & Animation: Tailwind CSS & Framer Motion

Audio: Native Web Audio API (MediaRecorder)

Backend

Framework: FastAPI (Python)

Data Layer: Supabase (PostgreSQL) / Internal Mock Database for dev

Inference Routing: OpenRouter API

Audio Processing: Whisper (Local or API-driven transcription)

Challenges we ran into

Definitely get our Agent up and running as we want to keep the software working at 0 dollars in costs

Accomplishments that we're proud of

The UI. I think it’s quite intuitive and heavily abstracted. It makes using our software extremely easy to pick up and has a lot of utility

What we learned

I learnt react much more deeply, and to a level I didn’t I could reach until now.

What's next for empower

Closing the gap in the shortage of labour in the hospitality space one agent at a time

Built With

• fastapi
• next.js
• python
• react
• typescript