Case study · Holiday rentals · Holiday flats, guest houses, inns
Two hours every evening on emails that almost always read the same.
An email assistant reads enquiries, checks the booking calendar and puts the finished offer on your phone for approval. You decide, it writes. The whole process is worked out here, with costs and limits.
Email work per week in peak season: two hours on seven evenings.
assumed · model business, 5 flatsAn evening at work
A family enquires about the autumn holidays. This is how the enquiry moves through the reference setup.
-
21:40The enquiry arrives.
Two adults, two children, 18 to 25 October, with a dog. The assistant reads along, classifies it as a booking enquiry and pulls out dates, people and requests.
-
21:41The offer is ready.
A free flat is found, the price is calculated by your rules, visitor tax and dog fee included. The draft lands in the cockpit and your phone alerts you.
-
21:44You approve.
A quick read, approve, done. Without your approval, no offer goes to a guest.
-
07:00The cleaning is booked.
The departure date turns into an order to your cleaning service, without anyone having to remember it. This does not even need the language model.
Example flow in the reference setup. The times show the order of events, not a measured response time.
Before and after, kept honestly apart
What was measured on the test setup is marked in green. What is assumed for the model business is dashed. We will recalculate with your own figures in a conversation.
Before
After
What the AI does. What you decide.
The assistant handles
- Read and sort enquiries: booking, question, damage report
- Pick out the details and check them against the booking calendar
- Draft offers and confirmations from your own text blocks
- Book cleaning and caretaker (add-on module)
You keep
- every approval before an offer goes to a guest
- prices, rules and exceptions
- complaints, special requests, anything that needs a personal touch
- the final say, always
The limit, stated openly
In the test, language models added details that were not in the email in difficult cases, in 1 to 3 out of 10 cases depending on the model. That is exactly why nothing goes out without your approval.
measured · 10 test cases per modelWhat is behind it
Four parts, nothing more. Your mailbox stays as it is.
Guests write to your address as before.
Runs on its own server, just for your business.
Only asked to classify an email and pick out details.
In the browser and on your phone: processes, approvals, contact person.
The server and the language model run on your own accounts with the provider. Expect around €10 a month for the server, depending on the provider, plus a few euros for the language model. You pay both directly to the provider.
Your cockpit. You see everything that happens.
No black box and no third-party platform you are tied to. Every process is visible, every approval traceable.
Real screenshots from my own cockpit instance. For you, the cockpit shows your modules, such as bookings and cleaning.
What it costs
Starting prices per component. The exact price depends on the number of your units, your special rules and the state of your data.
Email assistant with cockpit
- Read and sort enquiries
- Offer and confirmation drafts for your approval
- Cockpit with chat and push
- Support: technical updates and adjustments to your processes
- Trial run with your real enquiries and an introduction
All prices net plus VAT. With my guide, you book the server and language model directly with the provider: around €10 a month for the server, depending on the provider, plus a few euros for the language model.
A part-time helper on a mini-job basis (€603) costs you as an employer, with flat-rate contributions of around 31 %, about €790 a month. Source: Minijob-Zentrale, contribution rates 2026 (German).
The usual rate is 15 to 25 % of rental income (Favorent market overview (German)). Even at 15 %, that is around €750 to €940 a month for five flats. Worked example with €1,000 to €1,250 rental income per flat and month.
Want the numbers for your business?
Tell me how your enquiries are handled today, and I will work out with you what would be worth it. The initial conversation is 30 minutes and without obligation.
Book an initial conversationThe questions everyone asks
Where is my guests’ data stored?
On your own server. You set up the accounts with the host and the model provider yourself; they run in your name and under your contracts. I build the system and keep it technically up to date. The language model only receives the email text it needs to classify and read the message.
Which language model is used?
The default is IONOS, with data centres in Germany. On request, we can also connect OpenAI, Anthropic or Google, for example if a model works more reliably on your emails.
What happens if the AI gets it wrong?
You see it before anything is sent. Offers and confirmations are drafts until you approve them. If a detail is missing, the draft is meant to ask rather than guess. That does not always work, see above. This is why you read every draft before it goes out.
Can I leave again?
Yes. Support runs for at least ten months, after which it can be cancelled monthly with one month’s notice. The server and the model access belong to you: you keep them or cancel them under the terms of the respective provider.
How much work is the setup for me?
An initial conversation, your details on flats, prices, rules and text blocks, a trial run with your real enquiries and an introduction. How much time that takes depends mostly on what is already in writing, such as a price list, house rules and typical replies. We estimate that together for your business in the initial conversation.
Is this a real client?
No. The business is a model business with five holiday flats. The setup did exist, and technology, costs and setup effort were measured on it. The time saved is an assumption that we recalculate with your figures.
Let us work it out with your figures.
30 minutes, no obligation. You tell me how your enquiries are handled today, I tell you what is worth it and what is not.
Some images AI-generated