[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"glossary-prompt::en":3,"gloss-cluster-prompt::en":23,"gloss-next-prompt::en":9},{"slug":4,"category":5,"name":6,"definition":7,"meta_desc":8,"faq":9,"schema_markup":9,"related":10},"prompt","core-ai","Prompt","A prompt is the complete input given to an LLM to produce a response — it can be as simple as a single question or as elaborate as a multi-part instruction set combining a system prompt (persistent behavioral instructions), few-shot examples (sample input\u002Foutput pairs demonstrating the desired format), retrieved context (from a RAG pipeline), and the user's actual question. Prompts are the primary interface SaaS builders use to control LLM behavior without touching model weights, making prompt design (often called prompt engineering) one of the highest-leverage skills in building AI features — a well-structured prompt can turn a generic model into a reliable, narrowly-scoped tool for your exact use case. The core levers of a good prompt: being explicit about the desired output format (plain text vs. JSON vs. markdown), providing relevant context the model needs but doesn't have (your product's specific terminology, a user's account data, retrieved documents), specifying constraints (length limits, tone, what NOT to do), and — for complex tasks — giving the model room to reason before answering (chain-of-thought). A concrete worked example: a vague prompt like \"Write about our pricing\" will produce generic, unpredictable output. A well-engineered prompt for the same task looks like: `\"You are a pricing-page copywriter for a project-management SaaS. Write a 3-sentence summary of our Pro plan ($29\u002Fmo, unlimited projects, priority support, API access) aimed at freelancers deciding between Free and Pro. Tone: confident, no hype words like 'revolutionary' or 'game-changing'. Output plain text only.\"` — this consistently produces usable, on-brand copy because it removes ambiguity about audience, format, content, and tone. In production systems, prompts are typically templated (with variables like `{{user_name}}`, `{{account_tier}}` interpolated at request time), version-controlled, and A\u002FB tested like any other product surface, because small wording changes can meaningfully shift output quality, format compliance, and even hallucination rates. Prompt structure conventions have become fairly standardized across the industry: a system prompt sets persistent behavior and constraints (role, tone, output format, what to refuse); user and assistant messages carry the actual conversation; and increasingly, a dedicated \"tool\" or \"function\" message role carries the results of function calls back to the model. Builders should also treat prompts as versioned, tested artifacts rather than throwaway strings — a small wording change (\"summarize\" vs \"briefly summarize in 2 sentences\") can measurably shift output length, format compliance, and even factual accuracy, which is why mature AI products maintain a prompt library with change history and run regression evaluations before deploying a prompt change to production.","A prompt is the input text — instructions, context, and questions — sent to an LLM to elicit a specific output; prompt quality directly drives output quality.",null,[11,14,17,20],{"slug":12,"name":13},"few-shot-prompting","Few-Shot Prompting",{"slug":15,"name":16},"inference","Inference",{"slug":18,"name":19},"system-prompt","System Prompt",{"slug":21,"name":22},"temperature","Temperature",[24,28,32,36,39,42,45,48,51,54,57,60],{"slug":25,"category":5,"name":26,"updated_at":27},"agentic","Agentic AI","2026-08-24T02:46:36+00:00",{"slug":29,"category":5,"name":30,"updated_at":31},"alignment-tax","Alignment Tax","2026-08-24T02:46:37+00:00",{"slug":33,"category":5,"name":34,"updated_at":35},"artificial-intelligence","Artificial Intelligence (AI)","2026-08-24T02:46:38+00:00",{"slug":37,"category":5,"name":38,"updated_at":27},"attention","Attention",{"slug":40,"category":5,"name":41,"updated_at":35},"beam-search","Beam Search",{"slug":43,"category":5,"name":44,"updated_at":31},"benchmark-contamination","Benchmark Contamination",{"slug":46,"category":5,"name":47,"updated_at":31},"catastrophic-forgetting","Catastrophic Forgetting",{"slug":49,"category":5,"name":50,"updated_at":35},"computer-vision","Computer Vision",{"slug":52,"category":5,"name":53,"updated_at":31},"constitutional-ai","Constitutional AI",{"slug":55,"category":5,"name":56,"updated_at":27},"context-window","Context Window",{"slug":58,"category":5,"name":59,"updated_at":35},"deep-learning","Deep Learning",{"slug":61,"category":5,"name":62,"updated_at":27},"diffusion-model","Diffusion Model"]