Common questions.

REPLAI is an AI platform for guest communication in hospitality businesses, built in Austria. The plan is one inbox for WhatsApp, email, Instagram and Telegram, plus a guest app the guest opens via a link in the browser. The AI suggests replies, in your tone and in the guest's language, and you decide what goes out automatically and what comes to you for approval first. REPLAI is in development, beta coming soon.

REPLAI is meant for hospitality businesses of every size: the single holiday apartment on the side, the guesthouse with a team, the hotel and the property manager with 200 or more units. With one apartment it is usually about getting your evenings back, with many units it is about knowing who told which guest what. Built for the DACH region: Austria, Germany and Switzerland.

REPLAI is being rebuilt from the ground up. You can see a first demo, but no business uses REPLAI with real guests yet. First come the shared inbox, reply suggestions based on your property data (check-in, Wi-Fi, house rules, the area) in the guest's language, and the choice per property between approval mode and autopilot. WhatsApp, email, Instagram and Telegram are planned; messages from Airbnb and Booking.com run through your PMS. Nothing is fully connected yet. We deliberately don't name a launch date; you can follow the status openly under “Our path”.

You can see a demo, but testing with real guests isn't possible yet. Write to us and we'll show you the current state in an online meeting. For testing with real guests, join the waitlist and be the first to know when you can start. In the beginning you can test REPLAI for free. How long and to what extent, we'll agree with you in a conversation.

No, you don't need a PMS for the guest app, WhatsApp and email. All three are planned, none is finished yet. Messages from Airbnb and Booking.com, however, run through your business's PMS. Which system you use and what is possible with it, we clarify per business, so just tell us what you work with.

It doesn't guess, it hands over to you. Complaints, pricing questions or anything without the right context land with you as an open task. In approval mode you see every reply before it goes out anyway.

In approval mode nothing goes out that you haven't seen first. The AI suggests, you correct or approve. You define what the AI may write about at all, and anything outside those limits it hands over to you instead of guessing. When you correct a reply, you add to the knowledge for that property so the same mistake doesn't happen again. You set approval mode per property and channel.

As a rule, yes. Since 2 August 2026, Art. 50 of the EU AI Act sets a transparency obligation for AI systems that interact directly with people. REPLAI is therefore meant to make it visible when a reply comes from the AI. How the obligation applies in a specific case, for example to replies you approve yourself beforehand, is a legal question. This is not legal advice.

REPLAI is built by Timon Faux and Harald Fuchs, two co-founders from Lower Austria. The idea came from a night as guests when nobody answered; the product grows out of many conversations with hosts. We build in the open: what is done and what comes next is on the roadmap.

No. REPLAI Guest Communication from replai.at is a separate product from Austria and is not affiliated with Replai (replai.io) or Guesty's ReplyAI. The names sound alike, but different companies and products stand behind them. Here we always mean REPLAI Guest Communication at replai.at.

There is no price yet. We will only name one once we know our costs per property, and those depend on the rebuild. It will be shaped by the number of properties, the channels, your PMS and how much runs automatically. The only fixed point: no commission on your bookings. Those who help shape REPLAI early get better terms, and joining the waitlist is free.

The app's servers and databases are in the EU, and we are building REPLAI for GDPR from day one. Each business's data is meant to be kept separate, AI processing is meant to run in the EU, and guest data is not meant to be used to train models. Which providers we use will be listed in the product's privacy policy, which comes with the launch.