Writing · Nº 041065 essays
The writing
Ideas from inside the work. Explore technology, entrepreneurship, and the practice of building things that last.
36 essays found · Page 1 of 3
Clear filtersEngineeringThe Cost-of-Downtime Math Before You Over-Build DRDo the cost-of-downtime math before you buy redundancy. For most self-hosted apps, a simple restore beats an expensive hot standby you will never need.EngineeringHow to Handle Merge Variables in Email Templates SafelyA null merge variable sends 'Hi ,' to a real customer. Here is how to handle merge variables in email templates safely with defaults, guards, and escaping.EngineeringIs Self-Hosting Reliable Enough for Production?Is self-hosting reliable enough for production? Yes, if you do a few things right. Here is what actually drives uptime and why the managed premium is overrated.EngineeringAI Agents vs Traditional Automation: When to Use EachAI agents vs traditional automation compared: rules win on predictable tasks, agents win on judgment and messy input. How to pick the right tool for reliability.AIIs AI Bookkeeping Accurate Enough to Rely On?Is AI bookkeeping accurate enough to rely on? Accuracy is a property of the workflow, not the model. Here is how to measure it and what actually makes books reliable.EngineeringHow to Monitor Scrapers Before They Break SilentlyHow to monitor web scrapers so they fail loud, not silent: track real success rate, watch for soft blocks and data drift, and alert before a job dies quietly.EngineeringWhat to Actually Alert On When You Self-HostMost self-hosted monitoring alerts on the wrong things. Alert on symptoms users feel, not every metric, so the pages you get at 3am are always worth answering.ThesisAgent Reliability Beats Capability Every TimeThe thesis: AI agent reliability, not raw capability, decides which agents survive production. Why the dependable agent wins over the impressive one.EngineeringTreat Transactional Email as InfrastructureTransactional email is infrastructure, not a marketing feature. Here is why I treat it like a database or a queue, and what changes when you give it that respect.EngineeringHow to Make AI Agents Recover From FailureAI agents fail. Timeouts, bad output, dead APIs. Here is how to design agent failure recovery with retries, fallbacks, and escalation so a job never just dies.EngineeringSelf-Hosted vs Paid Uptime Monitoring: Which to UseSelf-hosted uptime monitoring like Uptime Kuma vs paid services like Pingdom. Why you should run both, and which one actually tells you your site is down.EngineeringHow to Test an AI Automation Workflow Before ProductionShipping an untested AI workflow straight to live data is how you break things at scale. Here is how to test automation with dry runs, fixtures, and a staging path.