Micheal's Substack
Articles about Micheal's Substack
Deconstructing the AI Self-Improvement Loop
Is the Machine Building the Machine? 5 Counter-Intuitive Truths About AI’s Recursive Future
How to Audit AI Predictions
Stop Panic-Reading AI Headlines: 6 Counter-Intuitive Truths About the Future of Work and Tech
From Weekend Project to Commercial Release
From Weekend Code to Commercial Reality: The Design, Legal, and Market Realities of AI-Assisted Game Creation
The Verifier Inside AI
Inside the Mind of an Agent: 4 Counter-Intuitive Truths About Monitoring AI
The Verifier Stack
5 Counter-Intuitive Lessons from Building Action Gates for Autonomous AI Agents
Check the Check
Why Your AI Agent’s Passing Tests Might Be Lying to You: 5 Hard Lessons from Landmark Machine Proofs
Can You Trust AI Confidence Numbers?
Why Your AI’s Confidence Is a Lie (And How to Fix It)
Latency vs. The Invoice: Benchmarking AI Decisions
The Speed Mirage: Why Your AI Strategy is Measuring the Wrong Clock
Jev in the Stack: High=Level Decision, Not Real-Time Control
Why Your AI Bot Shouldn’t Have Thumbs: Surprising Lessons from the Jev “Agentic Loop”
Grading AI Agents
How to Grade AI Agents When There Is No Test Suite: 4 Counter-Intuitive Lessons from the Cutting Edge
The Semantic If-Statement: Integrating AI into Code
The Death of the ‘Urgent’ Regex: 5 Surprising Lessons from the Semantic If-Statement
Give Your AI Agent Fast Reflexes
Why Your AI Agent Needs Reflexes, Not Just a Brain
Jev AI: Small Questions, Massive Piles
The Era of Decisions Too Cheap to Meter: 5 Surprising Lessons from the Jev Launch
Jev: The System One Decision Engine
Why the Future of AI Software is a “System One” Decision Engine, Not a Better Chatbot
From Prompting to Directing
Beyond the Prompt: 5 Lessons from the “Director’s Workflow” in AI Filmmaking
Effects That Belong: The Believability Blueprint
Why Your AI Video Feels “Fake”—and the One Science Fair Trick to Fix It
Writing the Edit: The New AI Filmmaking
The Ghost in the Machine: 5 Surprising Secrets to Editing AI Film Inside the Prompt
AI Filmmaking: Building Reusable Characters
Why Your AI Characters Keep Changing (and How to Stop It): 5 Lessons from the New Era of AI Filmmaking
AI Cinematography: Mastering the Move
The Death of Weightless Drift: Why Geometry is the New Prompt Engineering for AI Filmmakers
Build the Look: From Adjectives to Alibis
Stop Asking for “Cinematic”: How to Direct AI Like a Pro
Directing the AI Lens: One Shot, One Story
Stop Prompting, Start Directing: The Information Design of AI Cinema
The Two-Layer Harness: Surviving the AGI Shift
GPT-6 Astra and the 45-Point Gap: Why AGI Makes Half Your Code Obsolete (and the Other Half Priceless)
The Model Forgets, The Harness Remembers
The AGI Illusion: Why Your AI Remembers Nothing and Your Harness Everything
Govern Its Hands: The New Rules of Agentic Security
Why You Can’t Trust an AI’s Reasoning: A New Manual for Governing “Agentic” Loops
From Tokens to Tasks: The New AGI Invoice
The Invisible Invoice: Why Your AI Price Sheet is Lying to You
The Harness Decides the Score
The 37-Point Delta: Why GPT-6 Astra Proves the “Harness” is the New AGI Gatekeeper
Your Benchmarks Expired on Launch Day
Your AI Benchmarks Expired on Launch Day: Why You Need a Private Regression Gate
Decoding the GPT-6 Astra “Harness” Problem
The 37-Point Illusion: Why the “AGI is Here” Headline is Missing Three Vital Words