Articles
60 posts tagged "ai"
The harness is the craft.
Most engineers pick a model they trust, eyeball a couple of runs, and ship. I don't ship that way. Here's what eval-driven development actually looks like, and the seven principles I'd hand to anyone shipping LLM systems.
What Is a Claude Skill? Let's Look Closely at One.
A Claude Skill teaches AI to work the way you want, not the average way. Here's a walkthrough of one I built (Expert Profiler) and the install pattern you can reuse for any technical setup.
Why Your AI Bill Is Too High (And the Architecture Fix That Saves 40%)
Most companies run their AI like a doctor's office. The ones that scale run it like an ER. Here's the architecture decision that cuts AI costs by 40%.
The CRM I built this weekend doesn't have a UI. Agentic Software is the future.
I built a CRM this weekend with no UI. Just tools an AI agent can call. Here's why agentic software is the next decade of building.
You Vibe-Coded Something That Works. Here's How to Make It Real Without Touching the Code Yourself.
Vibe coding tools get you to a working draft in an afternoon. The next pass has a name and a shape. Five layers your AI can add for you.
AI Can Design a WordPress Site. It Still Doesn't Know Where the Design Should Live.
WordPress 7's AI features are real and they're going to help. But the hardest problem in WordPress isn't content. It's design. And it's an architecture problem, not a feature gap. Here's the two companies actually solving it for two very different users.
The Slop Isn't Coming From the Robot
Critics are right that most AI content is slop. They're wrong about where it comes from. The slop isn't the model. It's people typing one sentence into a blank chat window and pasting whatever falls out. Here's what separates the two camps, and the test you can run on anyone claiming to use AI well.
The Non-Developer's Guide to Building a Website with Claude
A 24-step walkthrough for non-technical readers — how to build, edit, and deploy a free website using Claude Cowork, GitHub, and Cloudflare Pages, without writing a single line of code.
The Two Bets Every AI Team Has to Make
Most AI teams celebrate the technical bet (prototype on the edge of working) without making the cultural bet that makes it survivable. You need both.
Your AI agents have one job each. Most builders give them four.
Most multi-agent systems fail for the same architectural reason: LLMs get wired into four positions when they only belong in three. Here's the rule that fixes it.
Want to talk about what you just read?
Every great conversation starts with a single question.