Notes on Building Right
Essays on human-centered AI, Behavioral Augmented Generation, and technology that gives professionals their presence back.
RSS FeedWhy "AI-Powered" on a Label Tells You Nothing — and What Does
"AI-powered" is now stamped on everything from software to toothbrushes. On why the label tells you nothing about quality, what actually separates a real tool from a thin wrapper, and the one question that cuts through the sticker.
Read the essay →Why Every AI Demo Looks Amazing — and What to Ask to See Past It
Every AI demo is impressive — that's the demo's job. On why a staged highlight reel tells you almost nothing about an ordinary workday, and the one thing to ask to see what the tool actually does when it isn't performing.
Read the essay →The 5 Questions to Ask Any AI Vendor — Before You Trust Them With Real Work
Every essay here ends with one question to ask an AI vendor. Here are five of the most important in one place — on data, grounding, uncertainty, workflow, and accountability — the checklist to keep open before your next pitch.
Read the essay →The Hidden Cost of "Free" AI Tools — and What You're Really Paying With
A free AI tool sends no invoice — but "free" software has always had a price. On what you really pay for free AI in data, control, and risk, and the one question to ask before you paste anything sensitive into one.
Read the essay →Why AI Won't Replace Your Best People — and What It Frees Them to Do
The anxious question behind every AI headline is "will it replace us?" On what AI is actually good at, the difference between a task and a role, and why your best people become more valuable, not less.
Read the essay →Why the AI Isn't the Hard Part — Your Workflow Is
The model is the easy part. On why AI succeeds or fails on how well it fits the tools and habits you already have, why a brilliant tool nobody opens is worthless, and the one question to ask before you buy.
Read the essay →Why a Bigger AI Model Isn't Always a Better One
The AI industry loves a bigger number, and buyers assume the biggest model must be the best. On why size isn't the same as fit, what a giant model actually costs you in speed, money, and privacy, and the one question to ask instead of "is it the newest?"
Read the essay →Why the Smartest AI Still Needs to Learn Your Business
A model can ace the bar exam and still be useless on your actual work. On the difference between general intelligence and knowing your business, what "grounding" means in plain English, and the one question that separates AI that fits your world from AI that just sounds smart.
Read the essay →Why AI Can't Always Explain Itself — and Why That Matters in Regulated Work
Sometimes an AI gives the right answer and still can't tell you why. On what "explainability" really means, why the black-box problem is disqualifying in law and medicine, and the one question to ask before you let AI near a decision you'll have to defend.
Read the essay →Why "Human in the Loop" Is More Than a Buzzword — and How to Tell If It's Real
"Human in the loop" is on every AI vendor's slide, but the phrase can mean real oversight or a rubber stamp nobody reads. On what meaningful human control actually requires, automation bias, and the one question that tells you which kind you're getting.
Read the essay →Why the Best AI Knows When to Stop — and Why That's Harder Than Being Smart
An AI that always has an answer feels helpful — until the day it confidently invents one. On why "I don't know" is hard for AI to say, why knowing its limits matters more than raw intelligence, and the one question to ask any vendor.
Read the essay →Why AI Vendors Get Vague About Your Data — and the One Question That Cuts Through It
Most AI vendors answer data questions with a reassuring line and a shrug. On where your information actually goes when you use an AI tool, what "we don't train on your data" leaves out, and the one question that separates real privacy from marketing.
Read the essay →Why AI Forgets What You Just Told It — and What "Memory" Actually Means
A capable AI assistant can lose the thread mid-conversation. It isn't broken — it's how these systems work by default. On the context window, what "memory" really means in plain English, and the one question that reveals whether a vendor built for continuity.
Read the essay →Why AI Pilots Stall — and the One Metric That Predicts It
Most enterprise AI pilots never reach production. The reason usually isn't the technology — it's adoption. On the pilot-to-production gap, why capable tools get abandoned, and the one number that predicts survival.
Read the essay →Why AI Sometimes "Makes Things Up" — and What Actually Fixes It
Every AI assistant will occasionally state something false with total confidence. It isn't a glitch — it's how these systems work by default. On why hallucination happens, what "grounding" means in plain English, and the one question that tells you whether a vendor has actually solved it.
Read the essay →Why Chatbots Feel Robotic — and What Behavioral AI Changes
Most chatbots give correct answers that still feel wrong. The reason isn't what they say — it's how. On the three behaviors humans expect in every conversation, and the one question to ask any AI vendor.
Read the essay →Why EMDELLE Exists: Presence Over Productivity
Every AI company promises to save you time. We started with a different question: what is that time for? On the conviction behind EMDELLE, and why "more human moments" is the test every decision has to pass.
Read the essay →