How I Work With AI

AI is fundamentally changing how we work. It boosts development output and makes it possible to solve problems of a complexity we haven't seen before. That's why I've thought deeply about my own workflows and knowledge architecture. The result is a development workflow in Obsidian that lets me handle most of my work template-based, agentically, or manually.

The context available to AI systems remains decisive for development quality. For AI systems to make meaningful use of that context, company knowledge should be structured generically and token-efficiently — for example via the Open Knowledge Format — so it can be used independently of the AI provider.

Structuring knowledge so it's easily accessible to AI systems also reduces the prep work needed for decision meetings, especially when information is generated or captured in a single source (single source of truth). This allows decision-making processes to keep pace with the increased development output. At the same time, the availability of full and historical context leads AI systems to deliver consistently better results over time, further increasing development quality.