Notes &
writing.
Short pieces on AI systems, custom software, and how we work with founders and companies. Written by the people doing the work, not a content team.
AIAgentsInfrastructureWorkflow DesignThe Agent Harness Is the Product Now
Anthropic's Managed Agents workshop makes a practical point concrete: real agent work depends on evolving harnesses, durable sessions, event logs, tool boundaries, context, and explicit outcomes.
May 30, 202611 min read
AIWorkflowsOperatorsSystemsJudgmentWhere the Human Should Actually Sit
The most important design decision in most AI systems is not what the agent does. It is exactly where, when, and how human judgment enters the workflow.
May 28, 20267 min read
AISkillsOperatorsWorkflowsStrategyThe Skills That Separate 3x From 10x AI Leverage
Prompting and basic tool use are no longer the differentiator. A new set of higher-order skills has emerged for the people getting real, compounding returns from AI systems.
May 26, 20266 min read
AISystemsWorkflowsMaintenanceOperatorsAI Systems Don't Stay Good on Their Own
The real cost of AI workflows isn't building them. It's the continuous, unglamorous work required to keep them from quietly rotting.
May 24, 20266 min read
AIAgentsDeveloper ToolsWorkflow DesignPi Is a Coding Agent for People Who Like Control
Pi clicked for me because it treats the coding agent as a configurable local system: models, MCP, themes, extensions, hooks, and skills all live close to the work.
May 22, 20268 min read
AIAgentsDeveloper ToolsWorkflow DesignThe Coding Agent Needs a Harness
Anthropic's large-codebase Claude Code guide makes a familiar point concrete: agents need context, tools, maintenance, and ownership around them before they can do serious work.
May 16, 20269 min read
AIAutomationSystemsWorkflowsThe Best AI Systems Are Event-Driven, Not Prompt-Driven
A lot of AI usage is still trapped in manual chat loops. The real step change happens when systems respond to events, schedules, and workflow state instead.
Apr 24, 20266 min read
AIStrategyToolsWorkflowsThe AI Stack Is Getting More Useful and More Confusing
Most people are not behind because they are lazy. They are behind because the AI stack is fragmenting faster than the explanations are improving.
Apr 14, 20266 min read
AICareersAutomationAI Won't Replace You. Someone Using AI Will.
Your biggest career risk right now is not AI — it's refusing to adopt it. Here's what's actually happening in production across 20+ businesses.
Apr 4, 20264 min read
AIAgentsWorkflowsStrategyStop Building Agents. Start Building Useful Capabilities.
A lot of teams are still overbuilding around the word agent. The real leverage usually comes from repeatable capabilities: workflows, context, tools, memory, and guardrails.
Mar 25, 20266 min read
AIAgentsAutomationProductivityWhat is an AI Agent, Anyways?
AI agent has become one of those terms that gets used so often it starts to lose meaning. Here's a cleaner mental model for what agents actually are and why they matter.
Mar 15, 20268 min read
AIMemorySystemsAgentsMemory Is Not Magic. It Is a System Design Problem
Useful AI memory is not about a model mysteriously remembering you. It is about deciding what is worth saving, where it should live, and when it should be written.
Mar 5, 20266 min read
AIProductivityClaudeHow Non-Technical Professionals Are 10x-ing Their Productivity With Claude Cowork
A practical guide to using Claude Cowork for real business automation — inbox triage, analytics, recruitment, and more. No code required.
Feb 23, 202616 min read
AISystemsContextStrategyContext Is a Better Upgrade Than a Bigger Model
Most teams still overfocus on model choice. In practice, better results usually come from better context, cleaner workflows, and stronger system design.
Feb 13, 20266 min read- AIEnterpriseAutomation
The Future of Enterprise AI: Beyond the Hype
How intelligent automation is reshaping the enterprise landscape — and why most companies are still getting it wrong.
Feb 3, 20263 min read