Services
Five services, named after the problem they solve.
I automate manual, error-prone processes – with Excel and VBA, Python data pipelines, dashboards, custom web applications and AI-assisted workflows. Which tool fits is decided by your process, not by my portfolio.
01
Excel and VBA automation
Spreadsheet work that eats hours every week and produces errors.
The problem
A team maintains reports, calculations or reconciliations by hand in Excel. The same steps every week, a high error rate – and only one person knows how the file really works.
The solution
I rebuild the workbook so the recurring steps run at the push of a button: cleanly structured data, VBA macros with error handling, clear inputs and outputs. The result is still Excel – your team doesn't have to learn anything new.
Typical deliverable
Automated workbook or macro-driven tool, documented and handed over.
A good fit when
… one team works in one file, the routine is clearly defined and the data comes from a few known sources.
02
Data pipelines & scraping
The data is scattered across portals, PDFs and systems – and gets pieced together by copy-and-paste.
The problem
The numbers you need exist – but in five sources, in different formats, behind logins or inside PDFs. Every week somebody copies them into a spreadsheet.
The solution
A Python pipeline collects the data automatically – from websites, APIs, files or databases – cleans it up and stores it, structured, where you need it: in Excel, a database or a dashboard. Repeatable, logged, maintainable.
Typical deliverable
Automated Python pipeline with structured output and operating documentation.
A good fit when
… the sources are known, the reconciliation is due regularly and the manual gathering is the bottleneck.
03
Data visualization
Decisions are being made on stale or unclear numbers.
The problem
The monthly report is ready once the month is already over. KPIs sit in three different files, and nobody knows which version counts.
The solution
An interactive dashboard or reporting layer wired directly to your data. Every metric is defined unambiguously, everyone sees the same state, and the numbers are current when you need them – not just at month end.
Typical deliverable
Interactive dashboard or automated reporting layer with documented metric definitions.
A good fit when
… the data exists, but preparing and distributing it is what slows good decisions down.
The problem
Several people work at the same time, versions circulate by email, access rights and history start to matter. The Excel file has become a liability.
The solution
A lean web application that does exactly what the process needs: database, user roles, traceable history. Built so it keeps running without me.
Typical deliverable
Purpose-built web application – for example a quote calculation tool the client now runs themselves.
A good fit when
… the process is stable and important enough to justify an application of its own.
05
AI-driven automation
Recurring tasks that call for judgment: writing reports, filing documents, sorting requests.
The problem
The task comes up constantly but can't be captured in rigid rules: summarizing texts, categorizing receipts, evaluating free text. So far a person does it – slowly and reluctantly.
The solution
AI-assisted automation that uses language models where rules no longer suffice – embedded in a controlled workflow with review steps, logging and clear boundaries. Not a chatbot experiment, but a tool with a defined output.
Typical deliverable
AI-assisted automation for report generation, classification or data extraction, with quality control.
A good fit when
… the task comes up often, the criteria can be described, and a person checks the result where it matters.
Frequently asked about services
Answers before you have to ask.
- What does an Excel automation cost?
- That depends on the scope of the process, which is why I don't quote flat rates. After a short description you receive a concrete quote with a clearly defined scope of work.
- Can you take over existing macros or tools?
- Yes. Many projects start with a file that has grown over the years and is first stabilized and documented, then extended.
- Who owns the code at the end?
- You do. You receive all sources, the documentation and a walkthrough so that your team or another provider can keep developing the tool.
- Do you use AI on sensitive data as well?
- Only with a clear data concept: which data does the model see, where does it run, what is logged? Where necessary, locally hosted models or anonymized data are used.
Next step
Which process is costing you the most time right now?
Describe it briefly. You will get an honest view on whether automation is worth it – and if it is, a concrete quote. There is no hourly rate and no flat fee on this site: effort and price follow the task. Some projects are done in a few focused hours, others run over several phases.