AI

AI

LLVM Creates "Human-in-the-loop" AI Policy

Updated: September 17, 2026

LLVM just published a new AI Tool Use Policy, and it's refreshingly boring in the best way: use whatever tools you want, but if you submit it, you own it. Read the output, understand it, test it, and be able to explain it in review. Label substantial AI help, don't let autonomous agents run wild in LLVM spaces, and don't "contribute" by dumping unvetted AI sludge onto maintainers. HITL isn't anti-AI. It's pro-not-wasting-everyone's-time. read on »

AI

Replit to Production: How I've Shipped 100+ Apps Without Losing My Mind

Updated: September 17, 2026

Replit is great for building fast. It's also great at reminding you that containers are temporary and your data is not. In this post, I walk through my boring (and therefore reliable) path from Replit to production: keep your database in AWS, wire Replit to GitHub, and deploy using date-based Git tags so you can answer "what's in prod?" without performing a ritual. It's simple, repeatable, and doesn't require the CEO to learn YAML. read on »

AI

AI Writes a Little Too Well, but Somehow Not Well Enough: The Linguistic Uncanny Valley of LLMs

Updated: September 17, 2026

LLMs write like a customer success email that's trying not to admit anything: clean grammar, tidy structure, a joke that feels like it was approved by Legal. And yet your brain still goes, "Something's off." This post digs into the linguistic uncanny valley-why model writing often sounds correct instead of true, why it avoids the sharp edges that make humans feel alive, and how the same instincts you use to spot fake "AI" and fake competence apply to text, too. read on »

AI

The SEO Paradox: How AI Ended the SEO Arms Race

Updated: September 17, 2026

You know what's ironic about the most recent AI revolution? It's actually pushing us to be more... human. I've spent decades watching the SEO world evolve from keyword stuffing to content farms to whatever the next "guaranteed page one ranking" trick happened to be. More recently, with AI churning out content faster than a caffeinated squirrel hoarding nuts, something unexpected has happened: authentic human expertise has become more valuable than ever. read on »

AI

What Are Tokens in LLMs? Understanding How AI Language Models Actually Work

Updated: September 17, 2026

I learned what a "token" was by watching an LLM ignore a simple persona and promote my $15k/year janitor to "CEO." That bug sent me down the rabbit hole of tokenization: why models don't read words the way we do, how BPE/WordPiece/SentencePiece slice text into chunks, and why "next-token prediction" explains both the magic and the lies. If you've ever hit a context limit or watched a model forget what you just told it, this is the real reason. read on »

AI

Vector Spaces Explained: Why They're Crucial for Modern AI, ML & NLP

Updated: September 17, 2026

Vector spaces: the unsung heroes of modern machine learning. From natural language processing to image recognition, these mathematical constructs form the backbone of AI. But what exactly are they, and why are they so crucial? This post demystifies vector spaces, exploring their history, basic concepts, and real-world applications. We'll dive into multi-dimensional visualization, compare traditional embeddings with LLM approaches, and even create a simple fruit-based vector space. Whether you're a beginner or a seasoned pro, understanding vector spaces is key to navigating the complex world of artificial intelligence. read on »

AI

Orchestrating Specialized Systems Trumps AGI

Updated: September 17, 2026

The pursuit of Artificial General Intelligence (AGI) may be misguided. Instead of creating a singular, omniscient AI entity, we should focus on orchestrating a symphony of specialized AI systems. This approach leverages current AI capabilities, allows for faster development and deployment, and mitigates ethical concerns associated with AGI. By embracing an AI orchestra rather than a single all-knowing entity, we can create practical, powerful AI tools that tackle complex real-world challenges while complementing human capabilities. read on »

AI

Why Old-School ML Models Still Win in Production: Naive Bayes, Logistic Regression & Decision Trees

Updated: September 17, 2026

Shipping "modern AI" inside a $200 gadget is a fast way to learn that benchmarks don’t answer support tickets. If your backend lags, customers don’t blame latency - they blame your device, then they ask for a refund. This post is a tour of the boring models that still survive production: Naive Bayes, logistic regression, trees, random forests, and even HMMs. They’re fast, cheap, debuggable, and they fail in ways you can actually explain. read on »

AI

The Third AI Future: Boring, Practical, and Quietly Useful | Small Models That Actually Work

Updated: March 18, 2026

Most AI futures are either killer robots or hoverboards. The one that shows up at work is boring: small models doing narrow jobs, under real constraints, and quietly cutting errors. This post is about the "third AI future" I keep betting on-tiny chips, local inference, measurable wins, and fewer wizard tricks (plus a confession about the ad tech era I'm not proud of). read on »