A custom solution for one company’s specific data flow
Three data sources arrive at one person, who every month copies, compares and recalculates them until the amount owed to each customer comes out. There is no system in the middle — there is a person, and that is exactly the place software can take over.
The situation
An electricity trader bought generated electricity from the owners of solar panels and wind turbines and sold it on through the exchange.
To prepare the monthly settlements, the company had to gather information from several sources:
- data on the electricity generated;
- exchange prices;
- customer and contract information;
- calculations of the amount owed to each customer.
A large part of the process was done manually.
In effect, a person was joining several systems together with their own labour:
pull data from one place → move it to another → compare → calculate → produce the result.
Why custom software was the logical answer here
A process like this is extremely specific.
It has:
- a fixed set of data sources;
- particular business logic;
- its own calculation rules;
- a particular customer structure;
- a particular end result.
There is no obligation to go looking for a universal product that tries to serve hundreds of different companies.
Sometimes it is far simpler to build a small system that does exactly one thing:
connect this company’s specific data and run this specific work process.
What we built
We developed a solution that automated a substantial part of this data-processing and settlement chain.
To do that, we had to understand:
- where each kind of data originates;
- how to join information held in different sources;
- how to identify one and the same customer or site;
- which calculations have to be performed;
- what to do with incomplete or incorrect data;
- what the end result has to look like for a company employee to work with it.
The software here did not replace some abstract “employee”.
It replaced one specific, repeatable flow of information.
Why a solution like this is especially realistic today
In the past, even a comparatively narrow internal system like this could mean a disproportionately large development project.
So companies often carried on working manually:
the process is not ideal, but building our own program looks too expensive.
Today that threshold has moved.
If the data sources are accessible and the business logic is understood, a first working version can often be built very quickly.
Which means custom software is no longer reserved for large systems or large budgets.
It can just as well be a small internal tool for one very specific company need.
What we would do first today
We would not start by developing the entire settlement system.
For €1000 we would take one real data flow.
For example:
pull the production data → pull the market price → tie it to a specific customer → run the main calculation → show the result.
The prototype could use real or anonymised company data, and would test:
- whether the integrations work technically;
- whether the data is of sufficient quality;
- whether the calculation logic has been understood correctly;
- how much manual work can be removed;
- where human checking is still needed.
Within a few days the company would already have a much clearer picture of what a solution of its own could look like.
If the prototype proves its value
Then the tested data flow can be developed into a full working tool.
A typical goal for the next stage would be 8–10 weeks to build a system that runs reliably day to day:
- receives data automatically;
- stores the information needed;
- performs the calculations;
- handles errors and exceptions;
- gives users the access they need;
- lets results be checked;
- behaves like a normal company system rather than a one-off script.
The main point
This example shows that custom software does not have to be an enormous platform.
Sometimes all a company needs is:
one connection between several systems, with its own business logic in the middle.
In the past a need like that was often easier to serve with a person’s manual work.
Today it is more and more realistic to build software for exactly that process.
And there is no need to make a big decision up front.
The sequence can be very simple:
a €1000 prototype → real data → we see whether it works → 8–10 weeks to a production system.
Do people in your company regularly join several systems together by hand?
Show us one specific data flow. Within a few days we can show how it could run automatically.