Writing
Thoughts on technology, DevOps, leadership, AI systems, and building great teams.
Personal Blog
Deep dives on the topics I care most about.
The Prototype-to-Production Wall
Most AI prototypes never reach production. The teams that cross the wall stop renting the model's guess and turn it into deterministic, owned capability - semantic to mechanical. The pattern, the data behind the failure rate, and how to cross it.
The Blameless Postmortem Playbook
Blameless culture is doctrine. The postmortem is where the doctrine survives or dies. A 60-minute agenda, a multi-perspective 5 Whys template, a facilitator script for the hard moments, an output template, and the follow-up loop that keeps the practice from becoming theater.
DORA Metrics for AI Agent Teams: A Reference Implementation
Most AI engineering teams measure model accuracy, eval scores, and cost per token. None of those measure delivery performance. The four DORA metrics translate one-to-one to AI agent teams once you redefine deployment - here is the operational framework.
Hiring for the Multiplier: The Interview Loop That Raises the Floor
Most hiring loops produce a thumbs-up or thumbs-down. A multiplier loop produces evidence. Three question archetypes, a written-first debrief, a reference check that surfaces peer-effect behavior, and a no rule that holds the floor.
The Multiplier Math: Why Team Composition Compounds
A great hire raises the ceiling 5-15%. A bad hire lowers the floor 30-40%. Both effects compound quarter over quarter. Team composition is the rank-1 leverage decision a leader makes - and most leaders spend most of their time on rank 4.
It's the Process, Not the Person: The Leadership Doctrine
Blameless culture is the most-cited and least-operationalized idea. We treat it as a leadership doctrine - one that shapes incident response, retention, hiring, and AI agent failure recovery - and show how to start practicing it tomorrow.
Collaboration Sessions: How Remote Teams Replicate the Best Practices of Co-Location
How to run effective collaboration sessions that replicate co-located team dynamics in a remote environment. Practical patterns from Team Topologies for stream teams, subsystem teams, and cross-functional daily collaboration.
Your AI Agent Works Great in Demo. Here's Why It Breaks in Production.
Every AI agent demo works. Then you deploy it. A field guide to retry amplification, silent degradation, cascading backpressure, recursive agents, and context bloat - the failure modes that separate demo-quality AI from production-quality AI.
Agentic Search Optimization: How to Get Your Brand Discovered by AIs
SEO optimized your website for crawlers. Agentic Search Optimization optimizes it for AI agents that browse, evaluate, and transact on behalf of users. A practitioner guide from someone who built the infrastructure on both sides.
How to Build an AI-Native Team: A Practitioner's Guide
Most AI-native team guides tell you to hire AI engineers and adopt Copilot. That is tool adoption, not transformation. An AI-native team is one where everyone has AI partners, the team is organized around Team Topologies, and AI agents are first-class members with governance. Here is how we built three of them.
Cross-Functional AI-Native Team Culture: From Silos to Shared Intelligence
Cross-functional teams already outperform siloed organizations. Add AI agents that carry context across disciplines - design to engineering, product to support, strategy to execution - and the advantage compounds. How to build the culture that makes it work.
Understand Before Acting: Why AI Agents Need Documentation Too
Your AI agent is the fastest coder on your team and the most likely to reinvent the wheel. A case study in why making AI read the README first prevents hours of rework.
MCP Is the USB Port for AI Agents
The Model Context Protocol solves the same N x M integration problem that USB solved for hardware. One standard connector, any AI client, any tool. Here is why MCP won, what it does not do, and what your team should build now.
One Governance, Many Orchestrators: CLAUDE.md, Cursor Rules, and AGENTS.md in Production
Everyone is comparing CLAUDE.md vs Cursor rules vs AGENTS.md. That is the wrong question. The right question is: where does your governance live, and how do you make every AI tool respect it?
The Autonomous Economic Agent Is No Longer a Whitepaper
In 2018, Fetch.ai published a whitepaper describing autonomous economic agents that transact, negotiate, and collaborate without human intervention. In 2026, those agents are running in production. Here is what changed and what it means.
From IC to AI Supervisor: The Leadership Role Nobody Trained For
Managing AI agents requires a new leadership skill set. Trust calibration, delegation frameworks, and oversight patterns for leaders building with AI.
Hire the Robots: Why AI-Native Teams Will Outrun Everyone Else
AI-native startups generate $3.5M revenue per employee. Traditional companies average $200-600K. The gap is widening - and it compounds. Teams that treat AI agents as team members will define the next decade.
The TCP/IP Moment: Why AI Agent Protocols Will Define the Next Decade
MCP gives agents hands. Chat protocols give them voices. Payment protocols give them wallets. Together they form a protocol stack that will define the next decade of software - and the teams that adopt early will have a compounding advantage.
Documentation Is Not Governance
An ETH Zurich study found that AGENTS.md files reduce AI agent performance. They tested documentation. We use governance. The distinction changes everything.
Vibe Coding Is Not Engineering
Vibe coding shipped fast and broke things. Engineering ships fast and keeps things running. The difference is discipline - and AI makes discipline more important, not less.
Social Intelligence Dynamics and Team Intelligence
Collective intelligence depends on social sensitivity, not individual IQ. New research on AI-enhanced teams, multi-agent architectures, and the social dynamics that make teams smarter than their smartest member.
Building Better Memory Systems for AI Agents
Why most multi-agent systems forget everything between conversations, and five architectural patterns that fix it. Entity disambiguation, trust boundaries, multi-participant sync, context filtering, and tiered storage.
The Zen of Frontend: Building a Culture of Idiomatic Code for Client Apps
How "Be Pythonic" and "Be Djangonic" inspired us to codify equivalent cultural principles for React, TypeScript, and Next.js. Eight aphorisms, community-sourced principles, and the cursor rules that enforce them.
Communication Is the API
What F1 pit crews, airline cockpits, and football playbooks teach us about engineering team communication. How to version your promises, lead with the "so what," and build communication infrastructure that scales.
Stand on the Shoulders of Giants: Why AI Agents Need Codified Principles
How a production bug caused by an AI-invented content format led us to codify a five-question decision framework that changes how teams govern AI-generated code.
Obey The Rules or Die: Enforcing Best Practices with Cursor Rules
Why writing rules isn't enough - you need iterative enforcement. How "Obey The Rules or Die" turns accumulated team wisdom into a compliance loop that catches what humans and AI agents miss.
Building AI Teams: Adversarial Agents and Role Specialization
Why teams of specialized AI agents with distinct responsibilities outperform monolithic assistants. Adversarial review, role specialization, and distributed context.
Spec-Driven Development with AI
Why planning before prompting delivers 6x better results than AI pairing alone. How we implement Spec-Driven Development for 10x productivity gains.
Why We Deploy Daily
How frequent deployments force better systems, faster feedback, and healthier teams. Why scheduled release days are an anti-pattern.
Cursor Rules for Teams: A Complete Setup Guide
How teams use Cursor rules to encode discipline-specific expertise while maintaining cross-functional awareness. Practical patterns for communication, persona, and shared team rules.
The AI-Augmented Team
How AI partners change the composition of cross-functional teams. Meritocracy means the best contributor wins - human or AI. Three levels of AI integration, practical team composition, and what leadership looks like when your team includes AIs.
When Multiple AIs Outperform One: A Developer's Perspective
How connecting Sentry AI, GitHub Copilot, and Claude creates debugging workflows that consistently outperform single-AI approaches.
Why Cross-Functional Teams Win
How team structure, clear objectives, and communication patterns determine whether your organization ships outcomes or just stays busy. Team Topologies, OKRs, and coaching anti-pattern language.
Getting Started with DevOps
A comprehensive guide to the fundamentals of DevOps: the 4 key metrics, CI/CD, cloud platforms, and security.
Getting Started with Pull Request Reviews
A comprehensive guide on what to focus on during code reviews. Help teammates develop solid habits and write maintainable, secure code.
You're Measuring Velocity Wrong
Why measuring features shipped isn't the right metric. Focus on velocity toward company objectives, not story points.
The Human Element
Dealing with the non-technical side of DevOps. Sometimes the answer is not another tool or script.
How to Facilitate Collaborative Coding While Remote
Strategies for enabling collaborative coding in remote teams. From git workflows to validation environments.
Release Branch Strategies
A guide to git branching workflows from single branch to GitFlow. Learn which strategy fits your team.