{"slug": "how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial", "title": "How to use LLMs effectively in your daily work — a practical tutorial", "summary": "A developer has outlined a practical methodology for using large language models (LLMs) effectively in daily software engineering work, emphasizing task decomposition and structured prompting over asking for entire features. The approach involves breaking work into tiny steps with one prompt per sub-task, giving the model specific roles and constraints, and using it primarily for review and critique rather than creation from scratch. The developer recommends treating different models as tools in a belt and adopting a \"reviewer-mode\" mindset where the human validates all AI output for correctness, security, and alignment.", "body_md": "**Exercise:**\n\nNext time you’re stuck, instead of “write this whole feature,” ask: “Which artifact is missing here (requirements/design/code/tests/docs)?” Then ask the model for that artifact only.\n\nSeveral prompt patterns are particularly useful for software engineering.\n\nGive the model a role, constraints, and an output format.\n\n**Pattern**\n\nYou are a senior {language/framework} engineer in a {domain} team.\n\nGoal: {one clear goal}.\n\nConstraints: {standards, stack, limits}.\n\nOutput: {bullet list, code block, checklist, etc.}.\n\n**Example (backend)**\n\nYou are a senior TypeScript engineer on a fintech backend.\n\nGoal: suggest a safe design for retrying failed payments.\n\nConstraints: Node 20, PostgreSQL, no external queues.\n\nOutput: brief architecture description, then pseudocode for the retry logic.\n\n**Exercise:**\n\nTake a current task and rewrite your next AI question into this pattern. Compare output quality with your usual “casual” prompt.\n\nBreak work into tiny steps: one prompt per sub‑task.\n\nInstead of:\n\nBuild a full auth system with JWTs in NestJS and write tests.\n\nUse a chain like:\n\nThis is “task decomposition” or “decomposed prompting.”\n\n**Exercise:**\n\nPick one real task you have this week. Write a list of 5-8 prompts that would get you from problem → working, tested code. Use them sequentially instead of one mega‑prompt.\n\nAsk the model to **attack** or **review** something, not to create from scratch.\n\n**Patterns**\n\nCode review:\n\nAct as a strict code reviewer. Given this code and these standards, list specific issues and suggested changes.\n\nDesign critique:\n\nAct as a system architect. Here is my design. Identify scaling risks, failure scenarios, and unclear responsibilities.\n\nGitHub’s Copilot guidance is essentially this: have AI review snippets and propose improvements; you still decide what to accept.\n\n**Exercise:**\n\nPaste one of your recent PRs (or a simplified version) and ask:\n\n“Act as a strict reviewer. What are the top 5 issues (correctness, readability, performance, security)?”\n\nThen decide which comments you’d actually address.## d) Self‑check / reflection prompts\n\nUse the model on its own output:\n\nRe‑read your last answer. Identify at least 3 potential errors, missing edge cases, or unsafe assumptions. Propose concrete fixes.\n\nThis leverages “introspective” prompt patterns shown to improve quality.\n\n**Exercise:**\n\nAfter a long answer you like, always follow with a self‑check prompt like above and compare.\n\nTrying to have an LLM build a whole feature usually produces fragile, untestable code. Instead, decompose tasks along artifacts and complexity.\n\nA blog on task decomposition shows exactly this: instead of “fix this whole feature,” it breaks debugging into focused questions like “What causes this error message?” and “What needs to change in auth.js?”\n\n**Example sequence (bugfix)**\n\n**Exercise:**\n\nTake a real bug. Before you touch it, use those four prompts. Compare your own diagnosis to the model’s; note where it helped or confused things.\n\nYou can treat different models as tools in a belt, picking per task.\n\nCommon usage patterns reported by practitioners:\n\nA typical “toolbelt” approach is to start with the fastest model to brainstorm or generate ideas, then move to a slower, more deliberate one to structure and refine.\n\n**Practitioner example:**\n\nOne architect uses Gemini wired into Claude so Claude can work over a large codebase, then uses Claude itself for heavier coding tasks and iterative refinement-but still doesn’t ask it to “design the complete product.”\n\n**Exercise:**\n\nWrite a small “LLM playbook” for yourself like:\n\nThen force yourself to follow it for a week and adjust based on what actually worked.\n\nThink of AI as a junior dev who can type very fast but is overconfident. Your job is to review.\n\nGitHub suggests validating that AI code compiles, runs tests, and matches your requirements and patterns. A broader “reviewer‑mode” checklist adds alignment, accuracy, completeness, and risk.\n\nAsk yourself:\n\n**Exercise:**\n\nTake one AI‑generated file. For each of the four dimensions above, write 1-2 concrete comments (like you would in a PR). Only then decide what to keep.\n\nThe risk with heavy AI use is “outsourcing thinking.” You can avoid that by **explicitly practising evaluation**.\n\nA 2026 article on “reviewer‑mode” suggests: strip away tone and just inspect the logic and assumptions. Ask:\n\n**Concrete habits**\n\n**Exercise:**\n\nPick one AI answer this week and spend 10 minutes writing a short “review” of it in a markdown file: what’s good, what’s wrong, what you’d change, what you learned. That’s how you turn AI usage into deliberate practice.\n\nReal‑world reports show successful patterns like:\n\n**Mini‑project exercise:**\n\nTake a small feature (e.g., “add rate limiting to this API”) and intentionally use AI in **each phase**:\n\nAt each step, you review and edit as if a junior wrote it.\n\nHere’s a compact set you can cycle through:\n\n**Decomposition drill:**\n\n**Prompt‑pattern drill:**\n\n**Model‑choice drill:**\n\n**Red‑team drill:**\n\nIf you tell me your main stack (e.g., “TypeScript/React”, “Java/Spring”, “Python/data”), I can turn this into a short, concrete “training plan” with stack‑specific example prompts and exercises.\n\nWhat tech stack and kind of work (feature dev, backend APIs, data, DevOps, etc.) are you mostly using LLMs for right now?\n\nRizwan Saleem — [https://rizwansaleem.co](https://rizwansaleem.co)", "url": "https://wpnews.pro/news/how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial", "canonical_source": "https://dev.to/therizwansaleem/how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial-3kdn", "published_at": "2026-05-29 22:09:18+00:00", "updated_at": "2026-05-29 22:42:50.036375+00:00", "lang": "en", "topics": ["large-language-models", "generative-ai", "ai-tools", "natural-language-processing"], "entities": ["TypeScript", "NestJS", "PostgreSQL", "Node"], "alternates": {"html": "https://wpnews.pro/news/how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial", "markdown": "https://wpnews.pro/news/how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial.md", "text": "https://wpnews.pro/news/how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial.txt", "jsonld": "https://wpnews.pro/news/how-to-use-llms-effectively-in-your-daily-work-a-practical-tutorial.jsonld"}}