Shubham Kumar Nayak
All writing

How I Built a School Utility App for My Mother

11 Jul 2025

EducationOCRAutomationPersonal ProjectWorkflow Design

A personal case study on turning school salary bills, leave summaries, official letters, OCR, translation, and remote printing into a practical workflow app for my mother.

How I Built a School Utility App for My Mother

Some software starts with a market opportunity. This one started with a phone call.

My mother was spending too much time on school paperwork: salary bills, leave summaries, forwarding letters, result summaries, and printed reports. Much of it depended on Excel templates, repeated formatting, manual checks, and remembering which document had to go where.

I first helped her manually. Then I realized the better answer was to build a tool she could use herself.


The real problem was workflow friction

The task was not just "generate a document."

The real workflow included:

  • keeping staff details ready
  • filling salary bill templates
  • calculating monthly values
  • preparing leave summaries
  • reading official letters
  • drafting forwarding letters
  • translating rough instructions into formal English
  • printing the final output

For someone who is not comfortable with Excel or IT tools, every small step adds stress.

That became the product goal:

Reduce the number of things she had to remember, edit, and verify manually.

Salary bills were the first useful feature

I started with salary bills because they repeated every month and had the highest chance of small mistakes.

The app reused the formats the school already trusted instead of forcing a new template. It kept a faculty master with salary-related fields, deductions, and editable values, then generated the final bill from structured data.

That decision mattered.

Good automation should respect the user's existing workflow when that workflow is tied to official formats.


OCR and summarization made letters easier

Official letters are often long, formal, and difficult to scan quickly.

I added OCR and summarization so she could understand the important points faster. The app could extract text from a document and produce a short summary with the action needed.

This was a small AI feature, but it was useful because it lived inside a real workflow.

The same thinking is behind my broader OCR and summarization pipeline.


Forwarding letters became a practical AI use case

Forwarding letters were another repeated pain.

My mother could describe what she needed in simple language, sometimes in Hindi or Odia, and the app could help produce a formal English letter with the right structure.

The important part was not the model. The important part was the shape of the output:

  • subject
  • recipient
  • formal body
  • amount or reference details
  • enclosures
  • closing/signature section

AI is most helpful when the product already knows the document format the user needs.


Remote printing completed the loop

Generating a document is not enough when the real world still needs paper.

The school printer was not cloud-enabled, so I used a Raspberry Pi with CUPS as a print server and connected it to the app workflow. That allowed generated bills and letters to print without making her handle files, USB drives, or printer setup every time.

This was a useful reminder:

In workflow software, the last mile matters.

If printing is still painful, the automation is incomplete.


What changed for her

The app helped her:

  • generate salary bills faster
  • prepare leave summaries with less manual checking
  • understand official letters more easily
  • draft forwarding letters in formal English
  • print documents directly from the workflow

The biggest change was confidence.

She did not have to wait for me for every small document issue. She could complete more of the work herself.


Engineering takeaway

This project taught me that meaningful software does not always start with scale.

Sometimes it starts with one real user, one repeated problem, and enough patience to understand the workflow properly.

The first version later grew into a larger system with multi-school support, employee records, AI-assisted letters, diagnostics, voice workflows, and remote printing. I wrote that continuation here:

Related: