Productivity on Pauls Blog

Sample data

Sun, 03 May 2026 00:00:00 +0000

The Strategic Case for Programmable Sample Data

I. The Core Thesis

Manual database entry is a bottleneck that introduces inconsistency. The most effective way to ensure environmental parity and development velocity is to treat sample data as code. By leveraging business services to seed the database with deterministic, UUID-based records, teams create a stable foundation for testing, UI development, and collaborative debugging.

II. Supporting Arguments

1. Enforcement of Business Logic via Service Seeding

A common mistake is seeding a database via direct SQL scripts. This bypasses the application’s validation rules and side effects.

Eliminate PR Noise with ktfmt

Tue, 21 Apr 2026 00:00:00 +0000

Few things are as frustrating during a code review as seeing a 50-line diff, only to realize that 45 of those lines are just re-ordered imports or adjusted indentation. This is “PR noise”.

In the Kotlin world, ktfmt is one of the best tools to solve this. Developed by Meta, it’s a deterministic formatter that ensures every developer on your team produces the exact same output, regardless of their individual IDE settings.

AI-Assisted Engineering: From Implementation to Specification

Mon, 20 Apr 2026 00:00:00 +0000

The more I use AI to assist with coding, the more I realize that context is the primary currency of effective assistance. The quality of the output is directly proportional to the quality of information available to the assistant.

This realization has fundamentally changed how I work. I’ve started treating ADRs (Architecture Decision Records) and design documents not as “after-the-fact” documentation, but as essential inputs to be versioned in Git alongside the code. Goodbye Confluence; hello documentation-as-code.