Zum Inhalt springen

Focus: local AI and Pseudomizer. Together. More

Industry partners wanted. Become a partner

EIGENKERNLocal AI
AI-generated

For hotels, spas and wellness

Twelve programmes that do not talk to each other.

Booking system, spa planner, staff rota, building services, till, robot fleet — each good in itself, none connected to the others. Energy is just the lever you see first. The more expensive problem is fragmentation.

The actual problem

It is not one system that is missing. It is twelve that do not talk to each other.

A wellness hotel runs booking, spa planning, staff rota, till, building management systems, CHP and increasingly a robot fleet — each tool decent in itself, none connected to the others. The result is double entry, handover errors and decisions for which nobody has the figures together.

The core reads read-only from the existing systems and answers in-house. No system is replaced, and none is given write access it does not have today.

How we connect

One translator instead of twelve individual interfaces.

The reason hotels shy away from IT projects is the interfaces: expensive, rigid, and due again with every update. We therefore do not program point-to-point connections but put a common translator in front (Model Context Protocol). The local assistant talks to it — and it talks to the systems.

ItemAssumptionFindings
Read-only firstEvery connection starts read-only.A system that only reads cannot disrupt operations.
In-houseGuest data is particularly protected.They do not leave the hotel network — the assistant runs on your hardware.
Step by stepNot a project that rebuilds everything at once.We start with one annoyance. Once that works, the next is added.

⚠ To be frank: we do not currently have a connection to a property management system in operation. What is written here is our approach — not a reference list. Plant technology, meters and load profiles, on the other hand, we connect on an ongoing basis.

The benefit

Ten things a learning assistant does for you.

Not what the technology can do — what it takes off your hands.

AI-generated01

Connecting the twelve programmes

A framework across booking, spa, building services and till — instead of twelve separate connections.

AI-generated02

Showing where the energy goes

Pools, ventilation, hot water, kitchen, laundry — separated rather than as a total.

AI-generated03

Finding the load peak

It usually arises somewhere other than expected — and determines part of your bill.

AI-generated04

Planning for the robot fleet

A fleet that charges simultaneously at night is a new peak. Can be shifted.

AI-generated05

Linking departure and cleaning

The room status is in the booking system — the robot does not learn of it today.

AI-generated06

Identifying existing technology

Before a second heat recovery system is planned on air that has already cooled.

AI-generated07

Securing knowledge for seasonal staff

House instructions as an assistant that answers in the native language.

AI-generated08

Ending the chain of Chinese whispers

A guest report at reception becomes a ticket with instructions — not a slip of paper.

AI-generated09

Cross-checking invoices

Against your own load profile, not against gut feeling.

AI-generated10

Providing evidence

With calculation path and source, for funding bodies, operators or certification.

Ready to use immediately

Tools that are already running.

No access needed, no registration — you can check the figures straight away.

From practice

An indoor and outdoor pool that wanted to actively control its systems.

The way in is almost always energy — because it is measurable and because the first finding is usually uncomfortable.

ItemAssumptionFindings
The planAn additional heat source from the backwash water should have covered the demand.It covered only a small part of it.
The ventilationA second heat recovery system had been considered.The first one was already there — the air at this point was already cold.
The real leverA new plant was being sought.What was found was the pool: it stores heat like a battery.
The patternThree assumptions, none of them malicious.All three fell apart as soon as someone calculated against the measurement data.

Reference case from an ongoing engagement, anonymised. Building physics and methodology, no operator data.

How we proceed

Three steps, in this order.

Each step delivers a result before the next begins. You can stop after any of them.

01

AI audit with benefit

We look at which data and documents you already generate today — and which of them are reliable. What you end up with is not a concept but a list: what pays off immediately, what later, and what you can save yourself.

02

Setting up and training assistants

The assistants are trained on your business — on your documents, your plants, your terminology. They keep learning the more often the same questions come up. Not a standard product that you have to adapt to.

03

Local database on suitable hardware

Your documents become searchable and stay in your premises — on a device matched to your data volume. No account with an American provider, no access from outside. Pull the plug: the assistant keeps running.

Next step

Look first, then calculate.

An initial conversation takes thirty minutes and costs nothing. Afterwards you know whether the first step is worthwhile for you — and if not, we will say so.

Bring us your biggest annoyance.

One conversation, one real task from your daily work. We use it to show what AI and automation can do for you.