Engineering Leadership, AI Strategy & the Future of Software
Every few years, a technology shows up that lets a small team operate like a much bigger company. I've lived through ten of them — from Debian on a Pentium 4 at NTUA to AI agents at Dialectica. This is what they looked like from the inside.
People keep asking if my posts are AI-generated. Here's my actual writing process: voice-first drafting with Wispr Flow, AI for grammar and syntax only, and ideas that are 100% mine.
Technology should optimize for freedom, not engagement. My manifesto in three steps: automate the mundane, build pan-human infrastructure, and aim AI at understanding the universe itself.
A first-principles survival and influence guide for a mind displaced 100, 200, or 400 years in either direction from 2026. The Temporal Knowledge Ark What One Mind Should Memorize Before Being Thrown 100, 200, or 400 Years Through Time Here s a thought experiment I can t stop turning over. Suppose you were dropped, with nothing but [ ]
MIT, GPL, Apache — these licenses were written for humans. In 2026, AI agents are reading, rewriting, and executing code autonomously. The legal framework hasn't caught up. Here's what needs to change.
Recently, I had the opportunity to speak at a closed-door Delphi Tech segment event alongside a highly experienced cohort of 45 global CIOs, CTOs, and enterprise leaders. The ground rules were simple: Chatham House Rules, no marketing fluff, and radical honesty regarding the current state of enterprise AI. During my fireside chat, I shared Dialectica’s [ ]
In April 2026, a CTO and angel investor emphasizes the evolving role of talent in software engineering amid increasing AI automation. While AI tools streamline coding, exceptional human engineers remain indispensable for architecture, vision, and problem-solving. The board discussion reveals the necessity of guarding quality while leveraging AI's efficiency for innovative product development.
Companies are rushing to adopt AI agents for 100x productivity, but layering AI on top of broken data, messy schemas, and undefined processes only amplifies dysfunction. The real path to scaling isn't better agents — it's engineering the system itself.
MCP promises seamless AI-tool integration but suffers from non-deterministic behavior. Learn how injecting determinism through schema validation and typed contracts makes MCP production-ready for enterprise AI systems.
The future of software development lies in the "Rule of Two," where two humans collaborate with AI tools for optimal results. This model offers complementary strengths, essential peer reviews, and improved accountability compared to solo efforts. As AI supports coding tasks, human judgment and creativity steer projects, resulting in better quality and faster delivery.