Pull instead of paste
Scheduled jobs that fetch from the source systems directly — an API, a database, an SFTP drop, or a mailbox that receives a daily CSV.
CASE STUDY
Example buildWeb AppBusiness Automation Dashboard
Replaces daily spreadsheet work with automated reports, alerts and role-based access.
- Type
- Scheduled jobs + dashboard
- Build time
- 8–12 weeks
- Platforms
- Browser, alerts to email or chat
- Starts at
- ₹80,000*
This is an example build. Tavzuran is a new studio with no client work to show yet. This page describes how the project would be scoped and built — the same thinking you would get on a real one. Nothing here is a past client, and no number on this page is a result.
The price follows the feature list. The number above buys everything scoped on this page. Take features out and it comes down — the shortest route to a smaller number is a shorter list, and deciding what to drop is what the free call is for.
THE SITUATION
What's actually broken.
Someone spends the first two hours of every working day exporting files, pasting them into a master spreadsheet, fixing what did not line up, and emailing the result to six people. That file has become load-bearing, and exactly one person understands it.
This is the most common software problem in a growing business, and the most under-rated one to fix. It is not glamorous. It buys back roughly five hundred hours a year and removes a single point of failure.
Two hours a day spent exporting and pasting
One person understands the master file
Six people read yesterday's numbers, not today's
Everyone sees everything, including margins
VERSION ONE
What we'd build first.
Every item here earns its place by being something a real user would miss on day one.
The rules, written down
Every fix that currently happens by hand becomes an explicit, tested transformation. The ones that cannot be pinned down get flagged for a human rather than silently guessed.
The dashboard
The numbers those six people actually read, live, instead of as yesterday's attachment.
Alerts
A message when a number crosses a threshold or when a source fails to arrive. Silence should mean everything is fine.
Role-based access
Finance sees margins, the floor sees volumes. Right now the spreadsheet shows everyone everything.
DELIBERATELY NOT IN V1
What we'd leave out.
Not because these are bad ideas. Because each one is designed better once the first version has told you something.
- Predictive analytics and forecasting. Get the reporting trustworthy first; a forecast on top of shaky data is worse than no forecast.
- Replacing the source systems themselves.
- A mobile app. Alerts go to email or WhatsApp; the dashboard is responsive and that covers the phone case.
UNDER THE HOOD
How it's put together.
- Scheduled jobs
- Python, which is the right tool for pulling and reshaping messy data
- Storage
- Postgres, with the raw pulls kept alongside the cleaned output so any number can be traced back
- Dashboard
- React with a charting library
- Alerts
- Email plus a WhatsApp or Slack webhook
- Hosting
- One small cloud instance and a scheduler
HOW IT RUNS
Week by week.
-
01
Week 1
Sit with whoever owns the spreadsheet and write down every manual step. This week matters more than the rest.
-
02
Weeks 2–6
Data pulls and the transformation rules, checked against the last three months of the existing file.
-
03
Weeks 7–10
Dashboard, roles and alerts.
-
04
Weeks 11–12
Both systems run in parallel and the numbers are compared daily until they match.
THE HARD PART
The rules nobody wrote down
Every build has one problem that decides whether the rest of it matters. This is that problem, written out.
The technical work here is ordinary. The hard part is that the spreadsheet contains years of accumulated judgement that exists only in one person's head: this branch codes its returns differently, that supplier's file has a blank first row, these two product names are the same thing, this number is always wrong on the first of the month and gets fixed by hand.
None of that is documented, and the person doing it often cannot list it on request — they only recognise it when they see it. So the discovery week is not a meeting. It is sitting beside them while they do the job and writing down every decision, including the ones they make without noticing.
What comes out of that gets split in two. Rules that can be stated precisely become code with tests. Rules that cannot get a flag: the system stops and asks a human rather than guessing. Guessing silently is how automated reporting loses people's trust, and trust here is the entire product.
BEFORE A PRICE EXISTS
What we'd ask you first.
A quote given without these answers is a guess. These are the questions that actually move the number.
- 01Which systems does the data come from, and can we get API or database access?
- 02How many manual corrections happen in a normal week, and what are they?
- 03Who signs off that the numbers are right today?
- 04What should happen when a source file simply does not arrive?
- 05Who is allowed to see margins, and who is not?
AFTERWARDS
How we'd know it worked.
Agreed before the build starts, so there is something to judge it against other than whether it looks nice.
- The two hours a day come back, and the person who owned the spreadsheet is doing something else.
- New numbers and old numbers agree for thirty consecutive days before the spreadsheet is retired.
- An alert fires before a human notices the problem, at least once.
- More than one person can run and explain the report.
BEFORE YOU START
Worth knowing.
Running both the old spreadsheet and the new system side by side for a couple of weeks is not wasted time. It is the only way to prove the new numbers are right before anyone trusts them.
Websites and apps start at ₹20,000*. Final pricing depends on features and project scope. Third-party costs — cloud hosting, domains, external libraries, APIs, app-store fees and any other paid service — are billed separately and are not included.
KEEP LOOKING
Other example builds.
Website + CRM
Real Estate Listing Portal
Property search, enquiry capture and a lead pipeline a sales team will actually use.
Read the case study Mobile AppDelivery Tracking App
Driver app with live location, proof-of-delivery photos and an owner-side status board.
Read the case study First VersionBooking & Scheduling, Version One
Ships in weeks, not months: online booking, reminders and payments for a local service business.
Read the case studyWant this built for your business?
Tell us what you're thinking. We'll figure out the next step together.
Free 20-min idea call · No obligation