Some reflections on my experience with LLMs and the AI landscape
No predictions, just a perspective from 25 years in tech, treating AI as another platform shift and a tool
Spike's Lab
Reflections on AI, PM, financial freedom, art, coffee and life
Live Last read · last published · next check
Latest posts
Prompts Are Not Workflows: rewriting Deep Research
Why I rewrote 2,000 lines of prompt into 6,000 lines of Python, and why the models getting better made it more urgent, not less
Claude Code Ate My Repo
How a race condition between AI sessions silently deleted 876 files
From Permission Hell to safe YOLO Mode
Containerizing Claude Code for autonomous, anxiety-free coding
How I built my deep-research skill
Eight Phases, Three Failures, and the Architecture of Trust
Thoughts and lessons learned from Going All-In with Claude Code
Focus on one thing and then iterate and refine and then iterate and refine some more.
On the Path to Autonomous Agents: Research Is the DIFFerence
What you don't already know is what you care about
On the Path to Autonomous Agents: Memory and Self-Learning
If we want useful agents we need to figure out a memory system
Output Styles: Claude Code's Hidden Feature for Non-Coding Work
Try it out and see the difference from a command or skill
I Built an AI Thinking Partner Because Critical Thinking is hard
100+ decisions a day and yet no school formally teaches how to think critically, isn't that absurd?