{
  "id": 18582562,
  "title": "Beyond code generation: A developer’s guide to agentic engineering",
  "url": "https://urgent.news/2026/11/06/beyond-code-generation-a-developers-guide-to-agentic-engineering",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-06T22:15:59.000Z",
  "source": {
    "name": "GitHub Blog",
    "slug": "github-blog",
    "url": "https://github.blog/ai-and-ml/github-copilot/beyond-code-generation-a-developers-guide-to-agentic-engineering/"
  },
  "original_language": "en",
  "account": "This October, GitHub experienced a record-breaking week with over 1.2 billion commits, a 400% increase compared to the same period a year ago. It's likely that developers are generating more code than ever before. However, simply generating more code isn't enough; developers need to consider other factors that contribute to the overall productivity of their software development process.\n\nFaros AI's study reveals that when teams transition from low to high AI adoption, pull requests spend 441% more time in review. This increased time in review can lead to delays and inefficiencies in the development process.\n\nAs a software developer, you are responsible for gathering requirements, steering implementation, reviewing changes, evaluating test results, and investigating issues in production. Fortunately, agents can help streamline the workflow across all these stages: Plan, Build, Verify, and Observe.\n\nIn the planning stage, it's crucial to ensure that everyone involved is on the same page. This may involve gathering scattered context from various tools, comparing different approaches, and agreeing on how existing clients will continue functioning during an API change. Copilot can help bring this scattered context together, work through tradeoffs, and turn decisions into an implementation plan. By using Copilot in collaboration tools like Slack or Teams, teams can efficiently plan together, assign tasks, and track progress.\n\nDuring the building stage, it's essential to continue progress without being tied to a single device. Session continuity allows you to pick up where you left off on another device, ensuring that work can continue seamlessly. Copilot's cloud sandbox execution keeps the execution off your machine, allowing you to return to a session later. This flexibility enables you to work on separate tasks without mixing edits, ensuring that each task remains focused and productive.\n\nIn the verification stage, developers need to spend their attention on high-value tasks. Reviewing pull requests can take a significant amount of time, especially when there are numerous changes spread across many files. Copilot can help manage this workload by allowing you to review changes more efficiently. By focusing on high-complexity work and delegating routine triage tasks to Copilot, developers can ensure that their time is spent on the most critical aspects of the development process.\n\nBy adopting an agentic engineering workflow using tools like Copilot, software developers can improve their overall productivity, streamline collaboration, and ensure that their code meets the necessary quality standards. With Copilot as a trusted partner, developers can tackle the challenges of modern software development with confidence and efficiency.",
  "summary": "Writing code got faster. The work around it didn't. See how you can plan, write code, review changes, and investigate production issues with GitHub Copilot, and what work still needs your judgment. The post Beyond code generation: A developer’s guide to agentic engineering appeared first on The GitHub Blog .",
  "key_points": [
    "GitHub saw 1.2 billion commits in October, a 400% increase from last year",
    "Faros AI study shows 441% longer pull request review time with high AI adoption",
    "Copilot helps plan, build, verify, and observe in software development workflow"
  ],
  "editors_take": null,
  "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."
}