↺ Back to builds

Internal Tool · Private

Database System

One vault, many windows.

One relational core replacing a 9-tab spreadsheet of 300+ contactsZero-duplicate by design: 'Add New' only unlocks after a search returns nothingThree tiers of access — Super Admin, Admin, Standard User — with maker-checker approvals

The Friction

Nine spreadsheet tabs — companies, contacts, venues, artists, F&B vendors, creators, agencies, projects, deals — holding 300+ relationships. Everyone dug through the same folders; nobody knew who spoke to whom last Tuesday; one wrong keystroke could delete the row that mattered. Institutional memory lived in individual chats and walked out the door whenever someone did.

The Catalyst

The architecture decision arrived when the operation committed to a Central Brain: a custom data layer that ingests, standardizes, and stores everything — because off-the-shelf platforms are built for technologists, not event professionals. The spreadsheet conversion became the seed of a bigger claim: every past and future event should land as clean, centralized data — data as capital, not exhaust.

The Resolution

A relational core where a Project tags its Venue, Artist, and F&B vendor once and the system connects the dots. Access runs on three tiers with an assignment engine — an admin hands a teammate exactly two brands to work, and that’s all they see. New entries wait in an approval queue before they exist; users correct their own logs with timestamped edits; one search bar finds the artist, the venue, and the manager who runs it. And the ownership rule that holds it together: the database owns itself — people get windows, never the vault.

The Pipeline

V1 discipline kept it lean: search-first entry so duplicates can’t be created, one CSV import to absorb the legacy spreadsheet, in-app notifications only, a mobile-first log flow for teams working from venues, forced reassignment when someone leaves, and an admin health screen that flags stale records at ninety days. The build-vs-buy fork was held honestly — spec is locked either way.

The Horizon

This is the Data Capital thesis in software: clean relationship history compounds into smarter planning for every future event — and it’s the precondition for the AI layer, which needs structured data before it can retrieve anything. The database stops being a CRM and becomes institutional memory that survives any single person’s exit.