In 5 weeks I built a system that replaces a shared Excel file from 2019.
TRS. Booking system for advertising walls, built for the sales team.
See the demo01Problem
Wall bookings ran in a shared Excel file from 2019. The file had worked for years, so nobody was looking for a replacement. I spotted the problem myself and brought it to the company. The company agreed to the change.
Assumptions
- Version 1 does one thing: a wall booking board instead of the spreadsheet.
- TRS connects to the CRM the team already uses.
- Each role has its own permissions.
- Every schema change is a migration in the repository, from day one.
- Long term, TRS is meant to become an external product.
02What I built
Booking board
A table of walls that shows at a glance what is free and what is sold. It replaces the shared spreadsheet.
Pipedrive integration
A connection to the sales team's CRM through its API.
Roles and sign-in
Sign-in through Clerk and separate permissions for each role.
Database with a change history
Supabase with a versioned schema. 173 migrations in 5 weeks.
Tests
127 unit test files and 67 e2e test files.
03Proof
- From scratch to v1
- 5wks16.09–20.10.2026
- Commits
- 1486in 5 weeks
- Migrations
- 173Supabase, versioned schema
- Test files
- 194127 unit · 67 e2e
04How
oohflow came first, a POC showing where TRS could go as an external product.
Then I built TRS v1. It is the smallest version that replaces the spreadsheet, a booking board with a Pipedrive integration. It took 5 weeks, from 16 September to 20 October 2026. I worked with AI agents and recorded every database change as a migration.
Next comes a roadmap toward an external product, modelled on oohflow.
- Time
- 5 weeks, 16.09–20.10.2026
- Tools
- Supabase
- PostgreSQL
- Clerk
- Pipedrive
- Unit and e2e tests
- AI coding agents
05Demo
Demo in progress
A working TRS instance on fictional data will appear here, with no real people or clients.