Analytics and AI leader George Andersen on making AI boring: methods for getting AI into production, and live projects that prove them.https://www.georgeandersen.neten-ushttps://www.georgeandersen.net/writing/why-i-built-this-sitehttps://www.georgeandersen.net/writing/why-i-built-this-siteFrameworks are claims until something ships. This site puts my methods next to the work that tests them, and dates my thinking so you can see it’s current.Tue, 29 Sep 2026 00:00:00 GMT<p>For my whole career, my job has been making complicated things clear: first in analytics, now in AI strategy and governance. This site is the same job, applied to my own work.</p> <p>It’s organized around one idea: <strong>theory, then proof.</strong> The frameworks are how I think about AI: what to do first, where it fits in the work, and how far to trust it. The projects are where I’ve tested those ideas in public. Each project shows the framework it puts into practice, and each framework lists the projects that apply it. Where a framework doesn’t have a public project yet, you can see that too.</p> <p>A few choices say something about how I work.</p> <ul> <li><strong>Every framework shows when I last reviewed it.</strong> After six months, the stamp turns amber. AI changes quickly, and ideas about it should be re-checked, not just published.</li> <li><strong>No cookies, no tracking.</strong> The site counts page views anonymously and nothing else, the same promise my projects make.</li> <li><strong>It was built with an AI assistant.</strong> I directed it, edited it, and checked it. That seemed like the right way to build a site arguing that AI should be useful, measured, and trusted exactly as far as it’s reliable.</li> </ul> <p>More soon.</p> Lab notesmetahttps://www.georgeandersen.net/frameworks/make-ai-boringhttps://www.georgeandersen.net/frameworks/make-ai-boringAI earns its keep in the boring part: in production, measured, and trusted exactly as far as it’s reliable.Tue, 29 Sep 2026 00:00:00 GMTFrameworkhttps://www.georgeandersen.net/frameworks/risk-tiered-adoptionhttps://www.georgeandersen.net/frameworks/risk-tiered-adoptionSequence AI by risk, not by one ROI gate: start inside, build the muscle, then go customer-facing.Tue, 29 Sep 2026 00:00:00 GMTFrameworkhttps://www.georgeandersen.net/frameworks/fix-first-ai-lasthttps://www.georgeandersen.net/frameworks/fix-first-ai-lastMap the work before you buy the tool: most stuck points need a process fix or a simple rule, not AI.Tue, 29 Sep 2026 00:00:00 GMTFrameworkhttps://www.georgeandersen.net/frameworks/calibrated-trusthttps://www.georgeandersen.net/frameworks/calibrated-trustFluency isn’t truth. Use AI where checking is cheaper than doing, and trust it exactly as far as it has earned.Tue, 29 Sep 2026 00:00:00 GMTFrameworkhttps://www.georgeandersen.net/projects/the-confident-machinehttps://www.georgeandersen.net/projects/the-confident-machineAn interactive essay on calibrated trust: why AI sounds sure when it’s wrong, and how to decide when to rely on it, check it, or keep the work.Sun, 27 Sep 2026 00:00:00 GMTProjectAI literacyinteractiveprivacy-firsthttps://www.georgeandersen.net/projects/what-do-i-actually-dohttps://www.georgeandersen.net/projects/what-do-i-actually-doA two-minute tool that maps your work week into a one-page blueprint: what to fix, what to automate, where AI helps, and what stays human.Sat, 26 Sep 2026 00:00:00 GMTProjectworkflowinteractiveprivacy-firstteams