AI Concierge

Answers your guests while you sleep.

A chat on your site that speaks four languages, knows your property, and recognises when an issue is urgent. Emergencies reach you by SMS — paperwork doesn't.

What it does

Multilingual chat per property

One link per property, embeddable on your site or via QR code in the home. Guests write in Italian, German, French, or English — the Concierge replies in the same language.

Knows your property

You upload FAQs, PDF instructions, links to the local guide. The Concierge indexes them and answers with precise references: hours, codes, emergency contacts.

Understands urgency

When the message is urgent (water leak, lock issue, medical emergency) you get an SMS in seconds. No alerts for routine requests.

How it judges urgency

Three levels, clear rules.

  1. 01

    Low

    Check-in questions, hours, nearby attractions. Concierge replies, no alert for you.

  2. 02

    Normal

    Practical requests (extra towels, parking). Email at end of day or in service hours.

  3. 03

    Urgent

    Issues that need immediate action. SMS to your number within 60 seconds, with a recap of the conversation.

AI providers

You pick, anytime.

The Concierge is model-agnostic. Switch providers from the settings panel without rebuilding the knowledge base.

  • Infomaniak AIDefault · Swiss-hosted, nFADP-compliant
  • Ollama (self-hosted)Open-source models on your VPS · no outbound traffic
  • OpenAI-compatibleAny compatible provider (Anthropic via Vercel, Mistral, etc.)
Privacy

Conversations stay here.

Guest chats are stored in your PostgreSQL database, under your control. History has a cap (8 messages for context, 600 tokens per reply) to keep AI provider costs predictable. No training on these conversations.

Let's look together.

A 20-minute demo to show how the Concierge is configured and how accurate it is on your FAQs.

Let's talk — 20 min