Writing
Notes on making AI work in real companies.
Essays from Haroon and Seeko on agent-native systems, AI readiness, and the operating discipline underneath useful transformation work.
AI-Native Engineering Is Boring (That's the Point) Everyone has access to Claude Code and Cursor. The gap between teams getting extraordinary results and teams getting mediocre ones isn't the tool. It's the boring infrastructure underneath. 10 min Five Futures for SaaS (and Which One Is Actually Happening) — Q1 2026 A quarterly look at five competing visions for the future of SaaS, why most of them are wrong, and what's actually playing out. Updated every quarter as the picture changes. 14 min Six Things I Keep Telling Every Company I Work With The same conversations keep happening across every AI engagement. Not the same questions — the same underlying mistakes. Here are six principles I find myself repeating in almost every first week. 7 min Most Companies Aren't Agent-Native. They Just Use AI. A framework for understanding where your engineering org actually sits on the agent-native spectrum, and why the jump from Level 2 to Level 3 is the hardest one. 7 min Agent-Native Systems Become More Deterministic Over Time, Not Less The counterintuitive truth about agent-native architecture: the best systems start flexible and gradually harden into mostly-deterministic code. The agent's job is to explore, then get out of the way. 8 min