It's 11:47 PM on Thursday. You have an apartment on Airbnb in Florence. The German guest arriving tomorrow messages you in chat, but you don't see it because you're already asleep: "Hallo, entschuldigung — ich habe Verspätung. Ich komme um 01:30 in Florenz an. Ist der Check-in OK?" Google Translate will read it for you in the morning, but the question needed an answer now.
That guest is going to arrive in Florence at 1:30 AM, in a city she doesn't know, looking for a door she has never seen. If she doesn't find precise instructions in her language, she'll try to call you. You won't answer. At 2:00 AM she'll leave a worried message. At 9:00 the next morning, when you wake up, the check-in will already have happened — but the mood of the review is already set.
This is the central problem for Italian short-term rentals and B&Bs in 2026: it's not the competition, not taxation, not bureaucracy. It's the fact that your guests speak ten different languages and arrive at any hour, and you are one person.
60% of your guests don't speak your language
According to ISTAT figures for 2025, Italy logged over 479 million tourist overnight stays, a new record. 60.5% of them came from foreign visitors. In the non-hotel segment — the one that includes B&Bs, short-term rentals, Airbnbs, agritourism — growth is even stronger: +6.6% in Q4 2025, more than double the pace of traditional hotels.
In practice: six out of ten guests at your place speak German, French, English, or Spanish as their first language. According to Italy's Ministry of Tourism, 14.8% of foreign visitors come from Germany, 13.2% from France, 7.5% from the UK, 5% from the United States. And that's just the top four countries.
The Italian short-term rental market has surpassed hotels in bed count: 3.2 million versus 2.2 million for hotels. Over 754,000 listings are active on Airbnb, managed by about 350,000 Italian hosts. ISTAT counts roughly 30,000 B&Bs, almost all run by single owners or couples. The sector is huge, but its backbone is people managing one, two, maybe four apartments — not chains.
For most of these operators the language problem isn't theoretical. It's daily.
The ten questions that always come
Anyone who has run an Airbnb or B&B for a season knows them by heart. What's the WiFi password. What time is breakfast. Where's the lockbox with the key. Can I check in before 3 PM. Is parking included. Are dogs allowed. What time does the metro close. Where should I eat tonight. Can you call a taxi for the airport. I can't find the front door.
They never change. They come at 10 PM, at 1 AM, at 6 in the morning. The problem is that every guest treats them as unique — they're asking them for the first time — and they expect a fast answer. Airbnb knows this and measures it: hosts who reply in under an hour convert far better than those who reply within 24 hours, and the listing's ranking is directly affected by average response time. Superhost status requires a 90% response rate within 24 hours, but hosts who want top positions respond in under an hour.
If you run a single apartment and have another job, you can't keep the phone in your hand all day. If you run four apartments and get thirty messages a day, you end up answering everyone in broken English, even when the guest wrote in French. The outcome is predictable: 4-star ratings on communication instead of 5, and a small but constant erosion of your ranking.
It's not just the language. It's also the hour.
The second half of the problem is that foreign guests come from different time zones, with flights at awkward times, often late evening or at night. A Frankfurt flight lands at Milan Malpensa at 11:40 PM. A New York flight arrives at Rome Fiumicino at 6:30 AM. The Tokyo traveler messaging you at 3:00 AM Italian time is wide awake, in virtual-office mode, finalizing a trip.
The result is that most operational requests — the ones that need a fast answer to avoid a problem — arrive in the hours when you're not there. After-hours analysis of small-business messages shows that about 40% of inquiries come in between 6:30 PM and 9:00 AM. For an Airbnb host the share is probably higher, because tourism runs on every time zone in the world.
The traditional fixes don't cut it. Airbnb's preset automated messages only cover the standard booking phases and don't answer real questions. Translators like DeepL help, but still require you to be awake and on your phone. The only genuinely scalable answer is a system that understands the question, knows your apartment, and replies automatically in the sender's language — without you pressing a button.
What an AI assistant actually does for a host
A modern conversational assistant, built on a language model, is not a rule-based bot. It doesn't offer a menu "press 1 for check-in, 2 for breakfast". It understands a question written in natural language, recognizes what language it's written in, and replies in that same language, drawing on information you've predefined.
The mechanism is simple. You upload your guide once: check-in times, lockbox instructions, the precise address of the front door, WiFi password, breakfast hours, pet and smoking rules, local dining tips, emergency number. The AI reads it all and uses it to reply. When the guest writes in German, the AI answers in German. When they write in Japanese, the AI answers in Japanese. It doesn't block-translate an Italian guide — it generates the answer directly in the guest's language, matching tone and context.
The system can live in three places, depending on how you work. A widget on your website (if you have a direct site outside Airbnb — many hosts do, to reduce commissions). A dedicated page at a link like chat.yourname.com, that you drop into the Airbnb or Booking confirmation and the guest saves for reference. A QR code printed in the welcome book inside the apartment, which the guest scans as soon as they walk in and uses to ask "how does the oven work" or "what's the nearest restaurant still open".
One important point: you always control the system prompt. You decide what the AI can and cannot say. If you want certain questions always routed to you — a cancellation, a refund, a delicate situation — you set that. If you want the AI to offer, at the end of every conversation, a link to leave a review after the stay, you set that.
Three scenes at the end of high season
10:34 PM on a Friday in August. A Dutch couple has just parked their car in the center of Matera and can't find your B&B's entrance. They write in Dutch on the dedicated page: "We zijn aangekomen maar kunnen de ingang niet vinden." The assistant answers in Dutch with three lines: a description of the green door to the left of the pharmacy, the lockbox code, the room number. You sleep. At 8:45 the next morning you check the conversations and see it's all resolved. Breakfast is served without tension.
1:12 AM on a Tuesday in November. A Japanese guest writes from their room: "すみません、シャワーのお湯が出ません。" The assistant understands this is urgent — shower with no hot water — and replies in Japanese with the boiler reset procedure you loaded into the guide, promising to contact you in the morning if the problem persists. The guest fixes it themselves. At 7:30 AM you see the log, call, apologize, offer a free half-day luggage storage. The guest is impressed: "You speak Japanese?". No, your page does. But it's as if you did.
Sunday afternoon, March. A German family of five is deciding where to book for the August holidays. They've found your listing and want to know if children under 3 pay, whether there's a high chair, how late the playground in the square stays open in summer. They write to your property's dedicated page in German. The assistant answers point by point, cites your rate for under-3s (free), confirms the high chair, adds that the square stays open until 11 PM in summer. It captures name and email because the family asks to reserve. The next morning you have three new pre-qualified contacts.
In none of the three cases did you reply. Yet the guest received precise information, in their language, in real time. The margin you gain isn't just time. It's the difference between a guest who arrives relaxed and one who arrives angry.
What to check before you turn one on
Not all AI assistants are equally mature. For a host or a B&B, five criteria matter — and they're different from what matters for a large hotel.
Automatic language detection is the first. Many chatbots on the market force the user to pick a language by clicking a flag. A Japanese guest at midnight doesn't hunt for the flag. They just type. The system has to recognize the language from the message and answer without being asked. Check that the provider covers at least 20 languages, including Asian ones — Japanese, Korean, simplified Chinese — and Slavic languages like Russian and Polish.
Document upload is the second. You've already written a guest guide, maybe in Google Docs or PDF. The system has to ingest it and use it as a knowledge base, without forcing you to rewrite everything in an editor. This is what's called RAG — retrieval-augmented generation — and for a host it's the difference between a five-minute setup and a five-hour one.
System prompt control is the third. You decide how the AI introduces itself, what it must never say, which requests must always be routed to you. For a host with very few interactions a day this customization may seem excessive. When apartments grow to three, four, ten, it's the only thing that lets you scale without losing your mind.
EU data residency is the fourth. Your guests send you names, arrival dates, phone numbers, sometimes document scans — all personal data. It has to be stored in EU infrastructure, with a data processing agreement that's GDPR-compliant. From August 2, 2026, the EU AI Act adds a requirement: the guest has to know they're talking to an AI. A good system communicates this with a clear opening message, unambiguous.
Price is the fifth. Today there are complete SaaS plans starting at €32/month, with a 14-day free trial and no credit card required. For a host running a single apartment it's less than the monthly cost of one turnover cleaning. For someone running ten, it's so little it doesn't even register.
Hospitality doesn't change. The threshold does.
A recurring fear among hosts is that an automated assistant makes hospitality impersonal. The experience of the first hosts who adopted them says the opposite. Automation doesn't cover the moment of welcome — when the guest walks into your home, meets your smile, tastes your breakfast. It covers the threshold: the hours between booking and arrival, where the question is always operational and the right answer is almost always the same.
Strip away the hundred repetitive questions, and what's left — the restaurant tip that's not in the guide, the chat about the city, the room switch for an anniversary — is the human part. Which is what the guest comes back for, leaves five stars for, and sends their friend to.
Hosts who adopt no automated system in 2026 are going to face a fork. Answering everything in real time, in four languages, 24/7, is physically impossible. Not answering means lower ratings and fewer direct bookings. The third option — answering slowly and roughly — is the one that leads to those 4 stars we mentioned earlier. None of the three is a strategy.
Want to see how an AI assistant would work for your Airbnb or B&B, able to answer in the guest's language even at night? Paste your site URL (or a link to your listing) on iperchat.ai and try it in 30 seconds — free, no signup.