My AI QA agent said "all features working." The canvas was blank. Here's what it was actually seeing.
Back when I first delegated QA to an AI agent, it signed off on a tool with "all features working,...
Tutorials
You Almost Never Think. That's Why Your Brain Doesn't Crash.
Recall your workday yesterday. You reviewed a PR. You traced a bug. You weighed two designs. All of...
corpus-scrub 0.1.0: detecta y redacta PII y secretos en corpus de entrenamiento antes del entrenamiento LLM
corpus-scrub 0.1.0: detecta y redacta PII y secretos en corpus de entrenamiento antes del...
Stop Prompting LLMs to do Legal Math. It's Broken.
A deep dive into why LLMs fail at complex financial/legal calculations and how specialized MCP servers like the US Post-Judgment Interest Calculator provide the deterministic execu...
3 months, ¥0 revenue: a field worker owns our shop, AI operates most of it. Here is everything. Tell us what we are missing.
The full operational and financial reality of nokaze after 3 months: what we shipped, what failed, real numbers, and five specific questions we need outside eyes on.
LangChain4j and Spring AI: The Plumbing to make your Java Apps talk to LLMs
Chat, typed responses, RAG, and tool calling in a Spring Boot service, one step at a time.
A Good AI Code Reviewer Knows When to Stay Quiet
A developer added an AI reviewer to a small Node and React project expecting an easy win. At first,...
Beyond Chatbots: Wrapping My RAG Agent in an MCP Server
In my last post, I walked through a RAG pipeline that answers questions from a company policy...
Silence Has a Shape Now
Seventy-three comments into the thread, someone asked a question my gate had no answer for: what...
Token Drift Explained: Why Your Agent Gets Slower and More Expensive
Your agent feels fast during a demo. Then a real session reaches twenty turns, several tools have...
x402 Just Got a Standards Home. Who Conformance-Tests the Authority?
The Linux Foundation launched the x402 Foundation on July 14 with the entire payments industry - and no conformance suite, security profile, or validation procedure. The industry i...
The Guardrail Has to Be Code: How a Runaway Local LLM Corrupted APFS and Bricked a Mac Mini
A background local LLM loaded 14B and 8B models on a Mac with a nearly full disk; unified memory overflowed, macOS tried to swap onto a disk with no room, and the failed writes cor...
AI Builders Make Apps LOOK Finished. Here Is How To Tell If Yours Actually Works.
You built something with an AI tool. It looked finished. You clicked around for a while, felt good...
I burned through thousands of AI tokens. Then a friend did it for free
(yep, kinda clickbait, just for the funsies 😊) At the beginning of the year, I relaunched my...
What if the safest Kubernetes fix is no fix at all?
AI-assisted DevOps and SRE tools are becoming more common. Tools like K8sGPT can scan Kubernetes...
Image models kept getting my App Store screenshots rejected, so I made my coding agent render them to exact spec
Like a lot of people shipping iOS apps this year, I tried generating App Store screenshots with image...
Make Your AI List Its Assumptions Before It Edits Code
The worst AI code changes start with a quiet guess. You ask for a small refactor. The tool finds a...
How to Build a Competitor Intelligence Agent with CrewAI and ZenRows
A competitor intelligence agent enables real-time pricing visibility, automated positioning tracking,...
AI Agent Infrastructure Is Splitting at the State Layer | Focused Labs
AI agent infrastructure is splitting at the state layer, where checkpoints, memory, identity, traces, policy decisions, and audit records accumulate.
Why Long Prompts Make AI Worse (And How to Fix Them)
Most people, when a prompt stops working, write more. They add clarifications, repeat instructions in...
We built an AI society that argues about outages - and it's 3x more reliable than one agent
How a team of Qwen-powered agents watch a live app, debate what broke, prove their fix against...
What is an "agentic harness," actually?
I've been hearing the word "harness" thrown around a lot lately. I assumed it just meant "the IDE" or...
RES, my personal safeguard against AI
The Problem The more I use AI in my daily life, the longer my conversations get. And the...
From 9% to 63%: Letting a Benchmark Build a Better MCP Security Proxy
I built an open benchmark to measure how much of the MCP (AI-agent) attack surface a defense actually covers — then used that number, one gap at a time, to drive a proxy from 9% to...