Mapping the 42 steps
We took apart the whole manual Excel process step by step and determined what can be automated and what must stay under human control because of the risk of silent errors.
CASE STUDY · AUTOMATION
From 42 steps in Excel, 5 remain
Cesta von helps more than 1,100 children in 36 excluded communities. Their progress out of generational poverty is measured in neurodevelopment, attendance and family income, and until now all the outputs converged manually in Excel.
Cesta von is a Slovak non profit organization fighting generational poverty. Through the field programs OMAMA, FILIP, ZEBRA and AMAL it helps families and children in more than 36 excluded communities. Its early neurodevelopment assessment methodology was validated in cooperation with the University of Oxford and social return is measured with the SROI framework.
- Industry
- Non profit / social work
- Scale
- 36+ communities, 1,100+ children
- Methodology
- Early neurodevelopment · Oxford, SROI
- Cooperation
- 2026
Challenge
Silent zeros in the reports that Excel never flagged.
The data analyst had to download outputs from four separate applications, run her own R scripts and assemble the Excel sheets by hand. The biggest risk was silent failure: because of a gap in a family name from the CRM, SUMIFS returned zero instead of an error, and 20 of 144 families had an incorrectly zero income in the reports.
ProblemExpectations
Outputs from 4 applications, custom R scripts and manual sheet assembly
One process: CSVs go in, a finished XLSX report comes out
SUMIFS returned zero instead of an error, 14% of families with a wrongly zero income
Data matching with automatic correction of inaccuracies
openpyxl shifted columns but broke chart references in the XML
Charts repaired directly in the XLSX XML tree
Sensitive data of children and families could not leave the computer
100% offline processing without the cloud
How we solved it
Four modules, one application
Progress measurement, Family income, Progress over time and CREDI flagging. Each module takes CSV exports from the CRM and the field apps and returns a finished XLSX report with charts.
pandas: cleaning and matching
The application automatically fixes inaccuracies in names, for example spaces from the CRM that once broke SUMIFS, joins the exports and recomputes family incomes including progress over time.
Intervention directly in the chart XML
When the standard openpyxl broke chart references by shifting columns, we unpacked the XLSX as a ZIP archive and rewrote the c:f coordinates directly in the XML tree. PyInstaller packed everything into a single .exe (46.5 MB) with no Python installation.
Python, Flask, pandas, openpyxl, PyInstaller
37 of 42 manual Excel steps are now done by an offline application in 10 seconds. The children's sensitive data never left the computer.
Michal GregušINLOGIO
The calculations do not run through an AI model but deterministic Python code, eliminating the risk of LLM computation errors. Where a silent error could lurk we deliberately kept a step for human review: the EU SILC statistics from the public source of the Slovak Statistical Office are still added manually by the analyst.
Under a strict anonymity regime it was not possible to send raw personal data to external cloud models. Reports are therefore produced completely offline. The proposed next step, an encrypted cluster for semi-automatic e-mail drafts, works exclusively with anonymized IDs and always with a human in the loop.
%
37 of 42 manual steps in Excel
< s
instead of tens of minutes of manual work
100% offline, no transfer of sensitive data
Project outputs
- An offline .exe application (46.5 MB) running on localhost without internet or a Python install
- 4 modules: Progress measurement, Family income, Progress over time, CREDI flagging of children below the 5th percentile
- Automatic generation of XLSX reports and charts in under 10 seconds
- Fixed income data for 20 families (14%) that had a wrongly zero income
- 1,946 lines of Python code handed over to the client
Instead of assembling spreadsheets the analyst now evaluates real social impact. Field OMAMA workers get an overview of children at developmental risk sooner, and the families whose income used to vanish into silent zeros finally see their true progress.
Want similar results?
Look at your processes through the eyes of automation. Tell us what slows you down, or book a 30-minute call right away.
Free
Find a use case for your business
Enter what your company does and AI generates 5 concrete ideas where automation saves you the most time. It takes a minute.
Generate 5 AI ideas
Ambitious teams
trust us
"We see Invelity as a team of professionals who are always willing and able to help quickly with solutions across many areas. Today we cannot imagine our digital marketing without this partnership."
Ready to streamline
your workflow?
Tell us about your process. We will help identify what can be automated and where to gain efficiency.
Let's go!