Articles
Why AI Development Teams Need to Quantify Code Drift
Why AI Development Teams Need to Quantify Code Drift
When AI ships code faster than anyone can review it, velocity metrics go vertical — and drift accumulates silently. The Production Drift Ratio is the first metric designed to make that cost visible.
When AI ships code faster than anyone can review it, velocity metrics go vertical — and drift accumulates silently. The Production Drift Ratio is the first metric designed to make that cost visible.
AI-assisted Coding is Trading Craft for Speed. Let's not repeat history.
AI-assisted Coding is Trading Craft for Speed. Let's not repeat history.
Two centuries ago, factory owners replaced skilled weavers with power looms because machines were faster. Profits went up. And the accumulated knowledge of generations of craft vanished overnight, leaving behind machines that broke and fabric with severe quality issues.
Two centuries ago, factory owners replaced skilled weavers with power looms because machines were faster. Profits went up. And the accumulated knowledge of generations of craft vanished overnight, leaving behind machines that broke and fabric with severe quality issues.
Huntley’s Loom Changed How I Think About Control
Huntley’s Loom Changed How I Think About Control
AI changed me from "author" to "approver." It doesn't have to be this way.
AI changed me from "author" to "approver." It doesn't have to be this way.
What is Drift? (And Why Every AI-Assisted Developer Should Care)
What is Drift? (And Why Every AI-Assisted Developer Should Care)
AI now writes code faster and more prolifically than any team can review it. The result is a new category of problem that most teams feel, but no one has clearly named. That problem is drift — and understanding it is the first step to controlling it.
AI now writes code faster and more prolifically than any team can review it. The result is a new category of problem that most teams feel, but no one has clearly named. That problem is drift — and understanding it is the first step to controlling it.
What is Design-Code Drift? Why Handoffs Aren't Roundtrips
What is Design-Code Drift? Why Handoffs Aren't Roundtrips
Every team that ships software eventually faces this same invisible problem: the design and the code start out in agreement, then quietly stop being the same thing. A roundtrip is what's supposed to fix it — but only if the loop actually closes.
Every team that ships software eventually faces this same invisible problem: the design and the code start out in agreement, then quietly stop being the same thing. A roundtrip is what's supposed to fix it — but only if the loop actually closes.
Why AI Buries Your Request Under Output You Never Asked For: The AI Distraction Stack
Why AI Buries Your Request Under Output You Never Asked For: The AI Distraction Stack
RLHF-trained models can produce impressive-looking output that satisfies the surface of a request (not the intent). Here's what you can do about it.
RLHF-trained models can produce impressive-looking output that satisfies the surface of a request (not the intent). Here's what you can do about it.
Anchoring Bias: How AI Doubles Down Before Backing Down
Anchoring Bias: How AI Doubles Down Before Backing Down
Eighteen images and eleven exchanges about rope physics—A masterclass in what happens when an LLM anchors on its first output and won’t let go.
Eighteen images and eleven exchanges about rope physics—A masterclass in what happens when an LLM anchors on its first output and won’t let go.
Looks right, fails fast: The AI coding trap
Looks right, fails fast: The AI coding trap
Part 2 of 2: The AI Coding Reality Series -- 7 gaps between AI code and production-ready software
Part 2 of 2: The AI Coding Reality Series -- 7 gaps between AI code and production-ready software
AI vibe coding was the easy part. Production is where it breaks.
AI vibe coding was the easy part. Production is where it breaks.
Part 1 of 2 · The AI Coding Reality Series - There's a moment every team experiences about six weeks after they start shipping AI-generated code to production: The demo went perfectly. The sprint velocity numbers were off the charts. What used to take a senior engineer three weeks took an AI-assisted developer three days.
Part 1 of 2 · The AI Coding Reality Series - There's a moment every team experiences about six weeks after they start shipping AI-generated code to production: The demo went perfectly. The sprint velocity numbers were off the charts. What used to take a senior engineer three weeks took an AI-assisted developer three days.
AI Governance and Human-in-Control: Lessons from the Anthropic Claude Code Incident
AI Governance and Human-in-Control: Lessons from the Anthropic Claude Code Incident
What a packaging error at Anthropic reveals about the moment we’re all in—and what we choose to do next.
What a packaging error at Anthropic reveals about the moment we’re all in—and what we choose to do next.
Next-Token Prediction: What Happens When AI Hedges?
Next-Token Prediction: What Happens When AI Hedges?
The other day, I'd asked my AI assistant to audit part of my own codebase. While reviewing the output, it gave me a response that startled me:
The other day, I'd asked my AI assistant to audit part of my own codebase. While reviewing the output, it gave me a response that startled me:
Bidirectional Design-Code Sync? The Design-Code Roundtrip That Isn't
Bidirectional Design-Code Sync? The Design-Code Roundtrip That Isn't
Figma, Claude, and Codex announced bidirectional design-code workflows. Here's what they actually shipped — and the pain it leaves completely unsolved.
Figma, Claude, and Codex announced bidirectional design-code workflows. Here's what they actually shipped — and the pain it leaves completely unsolved.
Reweaving: Solving Code Drift in AI-Assisted Development
Reweaving: Solving Code Drift in AI-Assisted Development
In the face of a rapidly expanding AI development landscape, we need a fundamental shift in how we think about building software; from asking, “How fast can we generate?” to “How do we maintain quality as we generate?”
In the face of a rapidly expanding AI development landscape, we need a fundamental shift in how we think about building software; from asking, “How fast can we generate?” to “How do we maintain quality as we generate?”
The AI Augmentation Principle: Great AI Tools Should Build on Human Ability
The AI Augmentation Principle: Great AI Tools Should Build on Human Ability
AI coding assistants are fast, confident, and eager to please. But speed without understanding isn't augmentation — it's automation with a smile.
AI coding assistants are fast, confident, and eager to please. But speed without understanding isn't augmentation — it's automation with a smile.
Vibe Coding and the Illusion of Superpowers
Vibe Coding and the Illusion of Superpowers
A little over a year ago, I was curious about all the hype around “vibe coding.” I’ve spent my decades-long career working in software and exploring new possibilities with this amazing tech. Okay, vibe coding, challenge accepted.
A little over a year ago, I was curious about all the hype around “vibe coding.” I’ve spent my decades-long career working in software and exploring new possibilities with this amazing tech. Okay, vibe coding, challenge accepted.
Stop Chasing AI Drift: It’s time to shift from "human in the loop" to "human in control."
Stop Chasing AI Drift: It’s time to shift from "human in the loop" to "human in control."
When we create a document and print it, we expect that what comes out of the printer will match what we intended. And that usually happens. In the world of software, however, design (what you want) and code (what you get) live in different worlds created by different individuals who are evolving the product at different speeds.
When we create a document and print it, we expect that what comes out of the printer will match what we intended. And that usually happens. In the world of software, however, design (what you want) and code (what you get) live in different worlds created by different individuals who are evolving the product at different speeds.
We're Building Fast with AI in the Wrong Direction: Why AI development without users fails at scale
We're Building Fast with AI in the Wrong Direction: Why AI development without users fails at scale
I started my career as a software developer. I built what I was sure was amazing software that people would love. I was so proud of it that I went to watch users enjoy how great it was in their work. What I found was humbling: they couldn't even see the feature I'd spent months on.
I started my career as a software developer. I built what I was sure was amazing software that people would love. I was so proud of it that I went to watch users enjoy how great it was in their work. What I found was humbling: they couldn't even see the feature I'd spent months on.