Why we answer on WhatsApp

Most travel companies have spent a decade moving customers away from humans. The logic is understandable: humans are expensive and don't scale neatly. The result is a support experience where the fastest way to solve anything is to find someone on Twitter who will escalate it.

We went the other way, for a slightly unromantic reason: it is cheaper to answer a question well the first time than to process the consequences of answering it badly.

What this looks like in practice

You message. A person who has been to the house replies. If you ask whether the pool is heated in January, the answer comes from someone who has swum in it, not from a field in a database.

If something goes wrong during your stay — and occasionally something does — the same channel works. You are not opening a ticket. You are messaging the people who are responsible for the house you are standing in.

Where the AI fits

There is an assistant on this site called Niva, and it is genuinely useful for the first pass: narrowing forty homes to four, checking dates, answering the things that are written down. It answers only from what we have actually documented, and when it doesn't know something it says so and hands you to a person.

That boundary is deliberate. An assistant that guesses about a cancellation policy costs a real person real money.