{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreic3xgotojdxqe7xh43psurvpz26yezltxvosao2jshkp5ulh7aw5e",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mmesqnoyjbk2"
  },
  "path": "/t/codex-is-replacing-efficient-primitives-with-slower-ai-workflows/1381430#post_1",
  "publishedAt": "2026-05-21T15:59:13.000Z",
  "site": "https://community.openai.com",
  "textContent": "I’ve been using Codex for real development workflows (game dev, frontend, UI systems, creative tooling), and I think there’s an important UX issue emerging with “agent-first” interfaces.\n\nI understand the vision behind Codex:\n\n  * multi-agent workflows,\n\n  * orchestration,\n\n  * autonomous tasks,\n\n  * async execution,\n\n  * long-running jobs,\netc.\n\n\n\n\nI also understand why OpenAI may not want to build “just another VSCode fork”.\n\nBut right now, Codex often feels like it removed extremely efficient developer primitives before AI became reliable enough to replace them.\n\nThe result is that many workflows become slower and more frustrating than in a traditional IDE.\n\nA few concrete examples:\n\n  1. Filesystem / Explorer\n\n\n\nA classic filesystem explorer is not an outdated abstraction.\nIt is an extremely efficient cognitive and visual tool.\n\nExpanding:\nassets → audio → sfx\n\ntakes less than a second.\nNo prompt.\nNo latency.\nNo interpretation.\nNo AI cost.\nNo ambiguity.\n\nBy comparison, asking an AI:\n“show me all sfx music files”\n\nintroduces:\n\n  * friction,\n\n  * latency,\n\n  * probabilistic interpretation,\n\n  * additional cognitive overhead.\n\n\n\n\nAI should augment filesystem navigation, not replace it.\n\n  2. Direct manipulation vs AI interpretation\n\n\n\nSometimes I ask Codex to … change a color.\n\nBut in a traditional IDE, I can simply:\n\n  * click the color square next to a hex code,\n\n  * open a native color picker,\n\n  * visually explore the full color space in real time,\n\n  * instantly see what feels right.\n\n\n\n\nThis takes less than a second.\n\nUsing an AI for this is often worse:\n\n  * slower,\n\n  * more expensive,\n\n  * less precise,\n\n  * dependent on interpretation,\n\n  * and it breaks creative flow.\n\n\n\n\nThis is a fundamental distinction:\ndirect manipulation vs probabilistic interpretation.\n\nFor many creative workflows, direct manipulation is superior.\n\n  3. Creative workflows need visibility\n\n\n\nIn game dev / UI / creative coding workflows, developers constantly:\n\n  * browse assets,\n\n  * inspect folders,\n\n  * compare files,\n\n  * preview images/audio,\n\n  * tweak values,\n\n  * use terminals,\n\n  * inspect diffs,\n\n  * navigate visually through project structure.\n\n\n\n\nRemoving these primitives compared to traditional IDEs like VSCode, Cursor, Windsurf or JetBrains products creates unnecessary friction in real-world workflows.\n\nThe filesystem is not just storage. It is part of how developers think.\nAnd access to the code is not just about writing everything manually. Sometimes it’s simply about quickly adjusting a value, tweaking a parameter, or changing a small detail directly.\n\nMy concern is that some current AI interfaces assume:\n“If AI can theoretically do something, it should replace the old tool.”\n\nBut many classic tools are already extremely optimized for human cognition:\n\n  * explorers,\n\n  * terminals,\n\n  * color pickers,\n\n  * timelines,\n\n  * node graphs,\n\n  * viewports,\n\n  * curve editors,\netc.\n\n\n\n\nAI is most powerful when layered on top of these tools, not when replacing them entirely with chat.\n\nWhat I (and probably many advanced users) would love to see is:\n\n  * multi-agent orchestration,\n\n  * async tasks,\n\n  * AI-native workflows,\n\n  * real time collaboration,\n\nBUT ALSO:\n\n  * integrated filesystem explorer (even a simple traditional sidebar with folders, subfolders, files, extensions, and expand/collapse arrows would already remove a huge amount of friction)\n\n  * embedded browser/devtools,\n\n  * lightweight document/media preview (not full Office-like applications, but simple built-in viewers/editors for common formats such as .txt, .md, .csv, .docx, .xlsx, images and audio files, to avoid constantly breaking workflow by switching between multiple external tools)\n\n  * direct visual editing,\n\n  * low-friction interaction.\n\n\n\n\nThe future probably isn’t:\n“replace all tools with chat”.\n\nIt’s:\n“deeply integrated human + AI collaboration.”\n\nRight now, Codex sometimes feels like some incredibly efficient primitives were removed before the AI became good enough to compensate for their loss.\n\nThanks for taking the time to read this feedback.\n\n_(English is not my native language, so I used ChatGPT to help me write and structure this feedback more clearly.)_",
  "title": "Codex is replacing efficient primitives with slower AI workflows"
}