Eating My Own Dog Food: How I Used the Framework to Write the Post About the Framework

2026-05-06 · Source: AI & ML – Radar · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Intermediate, long

Summary

Marc Millstone details his process for writing an article, "Don't Automate Your Moat," by applying its own framework for matching AI autonomy to business risk and competitive differentiation. He categorizes his writing tasks into four quadrants: Full Automation for citation mechanics, Collaborative Co-Creation for the AI Gateway example and build-versus-buy framing, Supervised Automation for counterargument drafting, and Human-Led Craftsmanship for the opening, defining dimensions, and the core framework with its evidence. Millstone emphasizes that AI handled mechanical assembly and accelerated execution, but human judgment was critical for design choices, verifying claims, and ensuring accuracy, especially with source material like the Knight Capital loss figure. He also highlights using Claude as a "brutal" critic to stress-test arguments and the extensive effort required to maintain a distinct authorial voice, distinct from generic AI-generated prose.

Key takeaway

For AI Engineers and AI Product Managers developing content or complex systems, you should apply a risk-based framework to determine AI's level of autonomy. Prioritize human-led craftsmanship for core intellectual property and critical verification, while leveraging AI for mechanical tasks and as a "brutal" critic. This approach ensures accuracy, maintains unique voice, and strengthens the overall argument or system integrity, preventing quietly broken outputs.

Key insights

Match AI autonomy to task risk and competitive stakes, reserving human oversight for critical, high-impact areas.

Principles

• Delegate tasks where being wrong is cheapest.
• Own tasks where being wrong costs argument or trust.
• AI can be a critic, but human judgment is final.

Method

Categorize tasks by risk and differentiation. Automate low-risk mechanics. Co-create with AI on recoverable elements. Supervise AI drafts for verification. Reserve human craftsmanship for core ideas, definitions, and critical evidence.

In practice

• Use AI for citation assembly and formatting.
• Employ AI to draft counterarguments for verification.
• Prompt AI with adversarial personas for critical feedback.

Topics

• "Don't Automate Your Moat" Framework
• AI Autonomy
• Human-AI Collaboration
• AI-Assisted Writing
• Source Verification

Best for: AI Engineer, Director of AI/ML, AI Product Manager

Related on AIssential

• The Architect's Guide to the AI Era • Luca Mezzalira & Teena Idnani • GOTO 2026 — Architects must integrate AI by augmenting existing practices, prioritizing human judgment for critical decisions and governance.
• How People Actually Use AI Agents — Real-world AI agent autonomy is shaped by human interaction and trust, not solely by model capability.
• Don’t Automate Your Moat: Matching AI Autonomy to Risk and Competitive Stakes — Unchecked AI autonomy in development creates "cognitive debt," eroding understanding and increasing risk and competitive vulnerability.
• Agents Are Not Enough: The Next Bottleneck Is the Human Framework — Human frameworks, not agent autonomy, are the next critical bottleneck for effective and accountable AI systems.
• A Framework for Measuring Appropriate Reliance on Set-Valued AI Advice — This framework is the first to formally measure appropriate reliance on set-valued AI advice in human-AI collaboration.
• No Human at the helm. No deal. — Autonomous agentic AI demands a dangerous "trust transfer" from human judgment to systems, necessitating human command and verifiable governance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI & ML – Radar.