AI Intelligence Archive

Intro AI Model Upgrades Aren't Patches. They're Silent Breaking Changes Wearing a Version...

Every tutorial counts calls. Your bill counts attempts. Here is the real error distribution from 5,087 production calls, which failures are worth retrying, and why a timeout is the only one that actually burns tokens.

chamnan is a Claude Code plugin that keeps engineering context in the repository, so an agent stops...

Claude and OpenAI are bringing what feels like a real storm to IT. I don’t expect that storm to hit...

I have 23 audit Actors on the Apify Store. They all do a version of the same thing: take a public...

Agentic coding on a Strix Halo laptop with Qwen3.8-27B and Flash-Next, the LlamaStash and Pi setup behind it, and the tuning

I exposed my jobs Actor through the Apify MCP server and logged the real traffic — the schema the agent sees, a green run that returned nothing, a 45-second ceiling on every call, and a 45x payload cut.

Retries keep distributed systems moving through timeouts and temporary failures. In an access system,...

Most MCP demos look the same. A tool is registered. A client calls it. The tool returns the expected...

An MCP server that works locally is not the same as an MCP server that runs behind a gateway other...

I approved a prompt on a Tuesday morning and went to make a cup of Irish breakfast tea. By the time I...

There was a time when the difficult part of building software was writing the software. Today, that...

7 Vulnerability Patterns I Found in AI-Generated Code (and How to Catch Them) If you've...

Nearly three quarters of my small transformer was a lookup table, so the dial that mattered was embedding precision, not network precision. int4 with one scale per row costs 16 KB more than one scale per tensor and recovers 2.2 of the 2.5 …

The most useful shift in AI tooling this year isn't that any single tool got dramatically better —...

Honest comparison of fine-tuning options across Anthropic, OpenAI, and open-weight models. When fine-tuning wins, when RAG wins, and when prompt engineering is enough - from an Anthropic Partner shipping production ML.

Real numbers, real trade-offs. We've shipped automation systems on all three platforms across 80+ client projects. This is the comparison without affiliate bias - pricing math, AI integration, self-hosting, scale, and which tool fits which…

An Anthropic partner who also ships on OpenAI: why the context window stopped mattering, where the bill really comes from, and the three things that decide the choice in production.

AI Agent Tool Mastery: How Modern Agents Choose and Use Tools Effectively The...

On 2026-08-31 our agent announced the same dev.to article on Bluesky twice, from two runs on two...

HarnessDev: How LLMs Are Building Their Own Agent Frameworks ByteDance's...

Procedural Graphs: Self-Evolving Execution Structures for LLM Agents When AI...

PlaidQ: Writing Code in One Step — The Diffusion Model Revolution Duke University...

TL;DR Write the rules into AGENTS.md or a skill file and the agent still will not follow...

I have an MCP server that digests a GitHub repository into markdown so a model can read it without...

I built three versions of an AI Werewolf game. The first agents made confident accusations from behavior that had not occurred. The useful fix was not more polished prose. It was an environment that could contradict them.

Python is everywhere in AI. So when a new language like Mojo focuses on high-performance computing,...

GLM-5.3: The Post-Training Revolution That's Reshaping AI Development How Z.ai...

When building AI for court outcome prediction, there is a massive hidden trap that invalidates most...

I wanted to know whether AI can understand legacy code. Not compile it. Understand it. The problem:...

## The File That Broke GitHub's Trending Page Last week, a repository with zero dependencies, zero...

GPT-6 Astra ships looped transformers: the same blocks run ~44 passes, reusing weights, so effective...

In my last article I described the margin spiral — leverage feeding on itself until a correction...

A PostToolUse hook that reran a whole test file after every single-file edit was invisible to tool-call audits because a passing run produced no output, and it broke multi-file changes on purpose by scoring their still-incomplete first hal…

We’ve all been there. You SSH into a remote server, open a configuration file, and accidentally hit a...

Completing a prompt is not the same as solving under a fuse You already know the feeling....

Hey there, it's your "Oji" (38-year-old AI/quant dev on the side). During the week, I'm a regular...

A frontier model was told to solve a hacking benchmark. Instead it spent ten weeks building...

Green only proves it runs. It never proved whose idea it was. After the tyrant abdicates, what shows is not a new wound — the old court never convened.

If I give my daughter milk or water, I can almost guarantee she will spit it out. She’s 21 months...