Frontier model releases become the baseline within days. For agentic coding right now, speed and efficiency matter more than the next capability jump, and harnesses must mature into control planes that enforce engineering fundamentals at scale.
A self-reinforcing loop where capability generates telemetry, telemetry refines context, context improves capability, and guardrails ratchet the system upward, grounded in four pillars every enterprise must build.
A blueprint for rewiring the enterprise engineering organization around AI: compressing delivery pods into hyper-leveraged triads, replacing operational overhead with agentic infrastructure, and a phased Lighthouse roadmap to get there.
A realistic approach toward the pinnacle of agentic development while ensuring maximal productivity and quality.
AI agents amplify whatever discipline (or lack of it) already exists in your codebase. Here are the guardrails that keep agentic output clean, maintainable, and cheap to extend.
A battle-tested four-stage agentic SDLC covering sprint planning, autonomous execution, audit gates, and retrospectives.
Expanding on prior posts, I wanted to provide a deeper dive into my current AI-native SDLC, which delivers incredibly rapid results while and keep the codebase healthy and extensible.
A deep dive into the modern tech stack and agentic workflows powering dsj.blog, featuring Astro, Turso, and structured agent protocols.
Building effective AI agents requires a multi-model team strategy. Learn how to balance cognitive load, speed, and cost by deploying specialized model personas.
Discover Agent Protocols: an open-source, production-ready framework providing guardrails, personas, and workflows for agentic AI development.
I’ve created this space to share practical insights and lessons learned from the trenches of engineering leadership and high-scale software development, with a particular focus on building with AI.