I'm particularly interested in anyone who's doing spec-driven or spec-centric development with AI coding tools like Cursor, Kiro, Claude Code.

Spec-Driven vs Spec-Centric: Why the Difference Matters linkedin.com/pulse/spec-dri…

I've been experimenting with Cursor’s browser capabilities - it’s great for testing core UI features. The interesting challenge comes with chat-based AI where outputs aren’t deterministic, so your test path can change every run.

How do you measure success when there's no single 'correct' conversation path? I've found that evaluating AI systems requires a completely different mindset to traditional observability. #LLM #Evals #AI #Observability linkedin.com/pulse/from-apm…

I chaired a virtual board meeting today. Attending were a CEO, CFO, CTO, CRO, CPO, and a risk-averse Non-Executive Director. None of them were human. Here's what I did and why. linkedin.com/pulse/i-chaire…

🚀 Just released JobFlow, a Claude Cowork plugin to help you find, apply, and follow up on jobs with less friction! 🎯 🔗 github.com/duncanapm/jobf… ✨ What it does: • Job searches • Application tracking • Tailored resume suggestions • Follow-up automation • Interview prep

I know a number of my past colleagues are finding the job market tough to navigate right now. That’s why I built JobFlow — a free Claude Cowork skill that helps automate searching, applying, and following up for a job. Let me know how you get on! 🔗 github.com/duncanapm/jobf…

Auto correct on my iPhone just changed k8s to n8n - a sign of the times! What next?