[{"data":1,"prerenderedAt":145},["ShallowReactive",2],{"\u002Fwriting\u002Fai-native-development":3,"\u002Fwriting\u002Fai-native-development-related":140},{"id":4,"title":5,"body":6,"description":117,"extension":118,"featured":119,"kind":120,"meta":121,"navigation":119,"ogImage":122,"order":109,"path":123,"publishedAt":124,"relatedWork":125,"seo":126,"stem":127,"summary":128,"takeaways":129,"technologies":134,"videoDuration":122,"videoId":122,"__hash__":139},"articles\u002Fwriting\u002Fai-native-development.md","What AI-Native Development Looks Like Inside a Real Business",{"type":7,"value":8,"toc":107},"minimark",[9,14,23,27,34,40,46,50,56,62,68,73,77,86,90],[10,11,13],"h2",{"id":12},"ai-native-is-a-workflow-not-a-claim","\"AI-native\" is a workflow, not a claim",[15,16,17,18,22],"p",{},"Saying you use AI tools doesn't say much on its own — the interesting question is ",[19,20,21],"em",{},"where"," they sit\nin the process, and where they explicitly don't. Here's the honest breakdown of how I actually use\nthem.",[10,24,26],{"id":25},"where-they-help-most","Where they help most",[15,28,29,33],{},[30,31,32],"strong",{},"Implementation speed."," Once a design is settled, tools like Claude Code and Codex can turn it\ninto working code — including tests and boilerplate — dramatically faster than typing it by hand.\nThis is the biggest, most consistent win.",[15,35,36,39],{},[30,37,38],{},"Fast iteration."," Cursor is where I do tight, in-editor loops — small changes, quick feedback,\nstaying in flow on a specific file or function.",[15,41,42,45],{},[30,43,44],{},"Research and second opinions."," Gemini is useful for exploring unfamiliar APIs or getting a second\nperspective on an approach before committing to it.",[10,47,49],{"id":48},"where-judgment-stays-mine","Where judgment stays mine",[15,51,52,55],{},[30,53,54],{},"System design."," Architecture decisions, data models, and failure-mode thinking happen before any\nAI tool touches the problem. This is the part that determines whether a system is maintainable six\nmonths later.",[15,57,58,61],{},[30,59,60],{},"Review."," Every AI-generated change goes through the same review I'd apply to a human\ncollaborator's pull request — often more scrutiny, because it doesn't yet have a track record in\nthis specific codebase.",[15,63,64,67],{},[30,65,66],{},"The decision to ship."," Tests passing and code looking plausible isn't the bar. The bar is the same\none I'd hold any production change to: does this behave correctly under the real, messy conditions\nof the system it's going into.",[69,70],"architecture-diagram",{":steps":71,"title":72},"[{\"label\":\"Design (human)\",\"detail\":\"Architecture, constraints\"},{\"label\":\"Build (AI-assisted)\",\"detail\":\"Claude Code, Codex, Cursor\"},{\"label\":\"Review (human)\",\"detail\":\"Same bar as any PR\"},{\"label\":\"Ship + monitor (human)\",\"detail\":\"Judgment, not automation\"}]","Where AI sits in the process",[10,74,76],{"id":75},"why-grounding-matters","Why grounding matters",[15,78,79,80,85],{},"The failure mode of AI-assisted work isn't usually \"the code doesn't run\" — it's confident-sounding\noutput that's subtly wrong. The fix is grounding: giving the tool real, verifiable data and\nconstraints instead of open-ended requests, and treating its output as a draft to verify rather than\nan answer to trust. I built a small, concrete example of this idea in the\n",[81,82,84],"a",{"href":83},"\u002Fwork\u002Fprojects\u002Fai-ops-recap-agent","AI operations recap agent"," project.",[10,87,89],{"id":88},"related","Related",[91,92,93,101],"ul",{},[94,95,96,100],"li",{},[81,97,99],{"href":98},"\u002Fwork\u002Fcase-studies\u002Fai-assisted-automation","AI-Assisted Business Automation"," — the full case study",[94,102,103,106],{},[81,104,105],{"href":83},"AI Operations Recap Agent"," — a small, concrete example",{"title":108,"searchDepth":109,"depth":109,"links":110},"",3,[111,113,114,115,116],{"id":12,"depth":112,"text":13},2,{"id":25,"depth":112,"text":26},{"id":48,"depth":112,"text":49},{"id":75,"depth":112,"text":76},{"id":88,"depth":112,"text":89},"What it actually looks like to fold Claude Code, Codex, Cursor, and Gemini into real ecommerce systems work — where the tools help, where human judgment stays non-negotiable.","md",true,"article",{},null,"\u002Fwriting\u002Fai-native-development","2026-05-18",[98,83],{"title":5,"description":117},"writing\u002Fai-native-development","A practical look at using Claude Code, Codex, Cursor, and Gemini as part of a real, accountable engineering workflow.",[130,131,132,133],"AI tools are excellent at implementation speed, not at owning system design","Every AI-generated change gets the same review bar as a human pull request","The tools you reach for should match the task — exploration, iteration, or multi-file builds","Grounding AI output in real, verifiable data is what makes it trustworthy",[135,136,137,138],"Claude Code","Codex","Cursor","Gemini","psq03nSGbEitqAbB56OmLn8k3h54IG7OUbfgoSwAzrA",[141,143],{"path":98,"title":99,"summary":142},"How Claude Code, Codex, Cursor, and Gemini fit into a real development process — not just code snippets.",{"path":83,"title":105,"summary":144},"A tool that turns commits, issues, tasks, and calendar events into a daily operational summary.",1785688079476]