Landline

Inspiration

Hotel guests often call the front desk for simple requests such as towels, toiletries, dining information, or local recommendations. When staff are busy, these calls may go unanswered or require several transfers. At the same time, many hotel rooms already have a familiar and accessible interface: the telephone.

Landline reimagines that phone as an AI-powered hotel concierge.

What it does

A guest speaks naturally to Landline and explains what they need. The voice agent understands the request, answers supported questions, and routes physical service requests to the correct hotel department.

For example, a guest can request additional towels, ask about hotel services, or find useful information near the hotel. The system turns the conversation into a structured request that staff can review and complete.

Landline does not pretend to know private reservation or operational information it cannot access. When a request requires hotel authority, the agent escalates it to a staff member.

How we built it

We used ElevenLabs to provide the real-time conversational voice experience, including speech recognition, natural voice responses, turn-taking, and background-noise handling through WebRTC.

The application converts each conversation into structured information such as:

Request type
Room number
Department
Urgency
Summary
Request status

Hotel staff can then review and manage these requests through the application. Tavily can supply current public information about the hotel and nearby services. Stay22 can provide public property data and alternative accommodation options when a guest needs a new booking.

Challenges we faced

The main challenge was deciding which information the AI could answer safely and which requests required a real hotel employee. We separated public information, hotel-controlled information, and private reservation information instead of treating every question as a general search.

We also had to design the experience for noisy hotel rooms while keeping the conversation fast and natural.

What we learned

We learned that a useful hospitality agent needs more than a natural voice. It also needs clear data boundaries, structured tool calls, reliable request routing, and a path to human assistance.

Landline demonstrates how hotels can improve guest response times while allowing staff to remain responsible for decisions and physical service.