Checking third-party assumptions
Every adopted key figure against what the plant actually does.
Otherwise your name is under someone else's figure.
For energy consultants and engineering firms
Your calculation has to hold up.
An expert opinion stands or falls with whether its assumptions hold up in reality. A learning assistant checks your figures against the plant's measurement data — before someone else does.
From practice
A third-party profitability calculation, checked for plausibility.
Not replaced by a second study — recalculated against the building's real measurement data.
| Item | Assumption | Findings |
|---|---|---|
| The assumption | The assumed seasonal performance factors of the heat pumps came from the best point on the manufacturer's data sheet. | In annual operation they are well below that. |
| The heat source | The planned additional source should cover the demand. | On paper it covered only a fraction of it. |
| The records | A heat recovery system should be replanned. | It was already there — the exhaust air there already cooled. |
| The side finding | Something showed up in the load profile that nobody had looked for. | After one month the midday PV dip was missing: an indication of reduced yield. |
Reference case from an ongoing engagement, anonymised. Methodological findings, no operator data.
The benefit
Ten things a learning assistant does for you.
Not what the technology can do — what it takes off your hands.
Best point ≠ annual value
Manufacturer figures are laboratory values. The assistant shows the difference.
The difference decides whether the payback holds.
Reading measured variables correctly
Grid import is not total load, and certainly not generation.
A mix-up shifts the entire balance.
Annual balance instead of a rough estimate
A design point in winter does not answer an annual question.
Otherwise the statement only holds on cold days.
Calculating the source balance
Does the heat source cover the demand — or only part of it?
The whole concept depends on this answer.
Identifying existing records
What is already there does not have to be planned again.
Saves the investment nobody would have needed.
Finding silent errors
Patterns in the load profile that no monthly report shows.
This is exactly what an audit checks.
Range instead of a single value
A range with reasoning holds up in committee.
A single attractive value gets attacked in committee.
Source for every figure
So that you can answer follow-up questions instead of fearing them.
Without a source reference, all that remains is your word.
Once trained, it stays
Your audit logic stays in-house and gets sharper.
Knowledge stays in the office — even when someone leaves.
Where it runs
Client documents do not leave your office.
Anyone who assesses third-party plants has other parties' operating data on the table — load profiles, contracts, invoices. The question of where these documents are when an assistant reads them is not a formality. That is why Eigenkern runs on a device in your premises.
The calculator is at your site
No account with an American provider, no transfer to a third-party cloud. Your clients' documents stay in your premises.
No address on the open network
The machine cannot be reached from outside. What goes in and who uses it is decided by you — not by a provider.
Calculation path and source reference for every figure
For every value, the derivation and the file it comes from. You can trace every statement back to its source — and show that in the conversation.
The unplug test
Disconnect the device from the internet. The assistant keeps answering. That is not a promise but something you can check yourself.
Ready to use immediately
Tools that are already running.
No access needed, no registration — you can check the figures straight away.
Funding database
Federal government and all 16 federal states in one overview — with funding rate, maximum amount and requirements. The programme pages on file are checked automatically for changes every day.
To the funding databaseHow 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.
Collaboration
Your clients get the funding. We build the technology.
You are in front of businesses with an energy audit obligation, load profiles and ISO 50001. That is exactly where demand for AI and digitalisation arises — and exactly what funding exists for, which hardly anyone has fully in view. We build the systems, but we do not sit at your table.
You bring the case
A client who wants to tackle digitalisation or AI. You know the business, the figures and the people.
You apply for the funds
Depending on the federal state and programme, a grant of 50 to 70 percent is possible. Our funding catalogue tells you which programme applies — and which has expired.
We carry out
The technical implementation: local assistance on the business's own server, connected to meters, folders and existing systems.
The funding catalogue
Checked daily against the primary sources.
Not a catalogue that is copied out once a year. A watchdog checks the sources daily; what expires is marked as ended instead of being silently carried on. 59 programmes from the federal government, all 16 states and the EU, from 112 checked sources.
The limitations are stated up front
For instance when third-party services are not eligible for funding or a programme requires the applicant's own development work. That decides between approval and rejection and is not found in any other catalogue.
Documented per programme
Rate, cap, term, application route — and the primary source so you can check for yourself.
Access on request
The catalogue is not a download. We set up access for you and get in touch briefly — the collaboration is clarified in conversation.
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.
We look at which cases are on your desk and where a recalculation brings anything at all. If it does not pay off for your office, we will tell you in the conversation — that is cheaper for both sides.
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.