Client project / Education
From assigning school programmes to preparing reports. Instructors verify the answers; the system handles transcription and the information needed for reporting.
Executive summary
How it works
01 / 06
The system assigns a programme to an instructor and adds it to their calendar. Dates, locations and programme details are kept together.
Assigned programme
Event added to the calendar
02 / 06
Students complete a paper questionnaire at the end of the programme. The instructor uploads the photos to its record.
Ready for digitisation
No need to type every answer again.
03 / 06
AI extracts responses and transcribes written answers. The results are prepared for the instructor to review.
Questionnaire processing
04 / 06
The instructor checks the extracted answers and corrects any mistakes. AI prepares the transcription; a person confirms its accuracy.
Transcription review
Human review remains part of the workflow
05 / 06
If the number of questionnaires differs from attendance, the system asks for an explanation. Missing records cannot simply go unnoticed.
Illustrative completeness check
Five questionnaires are missing. Explain the difference before submitting.
06 / 06
The system prepares analytics and information for the grants team automatically. The team uses these outputs to report how funding was used.
Outputs from processed data
Reports are prepared for the organisation to submit to the relevant funding body.
Simplified illustration · no client dataAutomatic walkthrough · select any step
Challenge
Linkin Sphere, z. s. runs school programmes on the internet, AI, cyberbullying and working with photographs. Students complete a questionnaire at the end of each session.
Instructors used to transcribe the paper responses by hand. Each additional programme meant more administration and more opportunities for errors.
I designed and built a system that assigns programmes to instructors, gives them the location, schedule and delivery details, and adds the event to their calendar.
After a session, the instructor uploads photos of the questionnaires. AI extracts the answers and transcribes the text. The instructor reviews the results, makes corrections and submits the programme.
The system compares related records. If an instructor enters 20 attendees but uploads only 15 questionnaires, they must explain the difference. The record contains both the answers and the reason the set is incomplete.
Accuracy remains the instructor's responsibility. Their work shifts from typing every answer to checking prepared data and resolving discrepancies.
Submitted data feeds automatically generated analytical reports. The system also prepares a report for the grants team to document the use of funding.
The grants team uses these outputs for submissions to public authorities or the Czech Ministry of Education, depending on the grant. The system prepares the reports; their formal submission remains the organisation's next step.
Instructors transcribed answers from paper questionnaires by hand.
They photograph the forms, review the transcription and explain any discrepancies. Reports are generated from the processed data.
Less than CZK 0.02 in AI processing cost per questionnaire. Refining the AI instructions brought the average processing cost down to CZK 0.01875 per questionnaire. This is the AI processing cost, not the total cost of running the application or reviewing results.
Approximately 35 minutes saved per programme, on average. Instructors spend less time on administration after each session.
Experience from this implementation suggests an approximately 95% reduction in processing errors. This is an indicative figure from practice, not an independently verified measurement.
The system is complete and has been handed over to the client. It connects programme administration, questionnaire digitisation, data checks and reporting.
“Mr Hůle developed an internal application for our team to digitise paper questionnaires using AI scanning. It saves our staff considerable time and administrative work by bringing several previously separate processes together. We particularly appreciate his fast communication, flexibility as requirements evolve, and willingness to keep looking for a solution until it works exactly as we need. This approach is why we are continuing to work with him on further stages of development.”Director, Linkin Sphere, z. s.
