AI 14 min read
My Agentic Web Development Setup Is Basically a `while True` Loop
A multi-model web development pipeline — research, design, implementation, review — eventually orchestrated by a boring Python while True loop.
#agentic-ai
12 posts
AI 9 min read
Before You Build an AI Harness, Become the Harness
Before building software around an LLM, manually provide the tools and context it would have. The interventions reveal whether the workflow deserves a harness.
AI 16 min read
AI Feels Faster. The Data Is Less Certain.
AI coding tools feel transformative, but the productivity evidence is mixed. Part 2: perceived speed vs measured outcomes, and why review is the bottleneck.
AI 16 min read
AI-Native Software Engineering Isn’t About AI
AI-native engineering isn't about how much AI you use. It's a development model where AI is a primary participant in building software, not an assistant.
AI 18 min read
The Future of Software Engineering Is an Operating Model
AI-native engineering isn't about removing humans. Part 5: how roles, teams, economics and accountability shift when implementation stops being the main work.
AI 19 min read
The Model Is Only One Part of the System
A useful AI workflow takes more than a better prompt. Part 3: specifications, context engineering, agent workflows, and the harness that enforces boundaries.
AI 18 min read
When Agents Scale, Coordination Becomes the Product
One agent can misunderstand a requirement; several misunderstand it differently. Part 4: coordination, multi-agent orchestration and verification at scale.
AI 16 min read
The Model Is Not the Development Environment
I spent 12 million input tokens on a small monorepo change. The problem was not the model. It was the development environment around it.
AI 12 min read
Your AI Agent Should Be a Folder
After testing agent frameworks, I found the best personal AI agent isn't a platform — it's a well-organized folder of Markdown, scripts, logs and state.
AI 9 min read
Human-in-the-Loop Is a Lie We Tell Ourselves Before Clicking "Approve All"
Human-in-the-loop sounds responsible, but tired humans rubber-stamping prompts isn't governance. Real answer: programmatic boundaries, human accountability.
AI 8 min read
I Tried Building My AI Agent With Existing Agent Frameworks. Then I Chose the Boring Deterministic Way.
After evaluating multiple AI agent frameworks, I found that deterministic automation with an LLM inside beats magic autonomy for real operational work.
AI 9 min read
Why Use LangGraph for Your First Agentic AI Project
LangGraph forces you to learn state, tools, routing and human approval — the agentic AI fundamentals that survive framework churn.
