Blog

LLM-written code tends to default to database locks to deal with concurrency. A small Go and PostgreSQL benchmark shows what that costs: last write wins served 1.5 to 3 times more users than the locking patterns.

An internal AI assistant — named after the Japanese 時 / とき (time, moment, era, opportunity) — autonomously resolved 32.5% of intervened engineering support requests over four weeks. A look at how it's plugged in, its quirks, and where it's actually helping versus where humans still own the work.

After repeated attempts at LLM workflow automation, a first real success comes not from smarter models but from scaffolding — small, scoped tools around the LLM that iterate, validate, and fit into existing workflows without a chatbox in sight.

~15% of engineering time goes toward support work. After attempting a self-evolving knowledge graph and failing, a more practical AI-assisted approach emerged — one that lets a single engineer handle triage 10 to 30 times faster.

Most ETL pipelines combine ingestion, change detection, and transformation into a single process. Learn why we're pulling them apart into three independent stages — and the early signs that suggest it might be the right call.

A reflection on how AI tools are reshaping how we communicate and create content online. Using AI to express ourselves is powerful — when used with intent — but we risk dissolving into the systems we use if we're not careful to remain present in what we produce.

A follow-up reflection on dismantling an AI orchestration system and rebuilding a practical, overhead-free workflow. This article explores the gap between what's theoretically possible with AI agents and what's practically useful today.

A personal reflection on the transition from using AI as a coding assistant to fully embracing AI-orchestrated development workflows. This article explores the mindset shift required to move beyond traditional software engineering practices and adapt to a new paradigm where AI agents handle increasingly complex tasks.

Toggl Engineering shares how they expanded their PostgreSQL logical replication CDC pipeline to support event-driven architecture and OLAP features. The team discusses challenges overcome and practical optimization techniques for handling near real-time data transformations across multiple transactional databases.

Toggl Track faced challenges serving analytical reporting needs alongside their transactional database. The engineering team implemented a PostgreSQL logical replication pipeline to stream data changes in near real-time, but discovered critical misconceptions about LSN offset tracking that caused data inconsistencies in production.