{
  "id": 20978275,
  "title": "I Don't Write Code Anymore: How I Develop With AI Agents",
  "url": "https://urgent.news/2026/11/21/i-dont-write-code-anymore-how-i-develop-with-ai-agents",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-21T01:29:28.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/techoptio/i-dont-write-code-anymore-how-i-develop-with-ai-agents-5535"
  },
  "original_language": "en",
  "account": "Two years ago, the author was writing thousands of lines of code annually. Now, they rarely code at all. While they still work on software daily, their first step is typically asking an AI agent for assistance. This shift began gradually, starting with autocomplete suggestions and evolving into full-fledged features. Over the past year, advancements have been remarkable. The author went from relying on AI as a coding assistant to integrating it into their entire workflow. The most significant change is not typing less, but working at unprecedented speeds, enabled by multiple agents handling different features and projects while the author reviews plans, answers questions, and decides on releases.\n\nThe author's workflow involves describing the desired outcome first, rather than describing the code to write. This approach allows the agent to inspect the project and propose an implementation plan without any prior instructions about which function to edit. The author emphasizes the importance of planning before building, using Cursor's Plan Mode for anything larger than a simple modification. The initial plan may require adjustments, and the author often saves these plans in Markdown for future reference. For small fixes, the author skips Plan Mode and lets the agent make the edit directly.\n\nFor larger features, the author breaks them down into smaller issues, creates implementation plans for each, and lets agents work on them separately. This approach prevents overwhelming the author with one massive task. The author also uses Git worktrees and separate development server instances to ensure that each agent's changes and tests do not interfere with one another. This system allows for parallel development, significantly reducing waiting times and making it easier for the author to review, test, and catch mistakes in smaller pull requests.",
  "summary": "A couple of years ago, I was writing thousands of lines of code a year. Today, I don't really write code anymore. That sounds more dramatic than it is. I still work on software every day, and I still need to understand what the code does. But whether I'm implementing a large feature, changing one line, investigating a bug, or configuring infrastructure, my first move is usually to ask an AI…",
  "key_points": [
    "Author used to write thousands of lines of code annually.",
    "AI agents now handle coding, planning, and decision-making.",
    "Workflow involves describing outcomes, using multiple agents."
  ],
  "editors_take": "Shifting from manual coding to overseeing multiple autonomous agents transforms the developer role from a writer of syntax into a high-speed architect of implementation plans and project reviews.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}