The practice of designing and refining instructions given to AI models to get more accurate, useful, and reliable outputs.
The way you phrase a request to a large language model can significantly affect the result. Prompt engineering involves giving clear instructions, relevant context, examples of the desired output, constraints on format or length, and a defined role or audience.
For example, rather than asking an AI to write about cloud computing, a better prompt might ask for a 300-word explanation of cloud computing for small business owners in Nepal, with one real-world example and no jargon. Techniques include few-shot prompting, which provides worked examples, and asking the model to reason step by step. As models improve, some tricks matter less, but clear communication, good context, and testing prompts on varied inputs remain valuable skills.
A software system that uses an AI model to plan and carry out multi-step tasks, often by calling tools, APIs, or other programs on its own.
Chain-of-Thought PromptingA prompting technique that asks a language model to work through a problem step by step before giving its final answer.
Few-Shot PromptingGiving a language model a handful of worked examples inside the prompt so it can follow the same pattern for a new input.
Large Language ModelAn AI model trained on vast amounts of text to understand and generate human-like language, code, and other content.
System PromptBackground instructions given to a language model by the developer that set its role, tone, rules, and limits for every conversation.
Zero-Shot LearningThe ability of an AI model to perform a task or recognize categories it was not explicitly trained on, without any examples.
Sample prompts for getting real budgeting help from AI chatbots, where they fall short, and the financial data you should never paste into a chat window.
A no-code walkthrough for building an internal AI assistant with instructions, knowledge files and guardrails, including what OpenAI's switch from GPTs to plugins means for you.
Reasoning models think before they answer, which makes them slower and pricier. Here is how they work, what they cost and when the extra thinking pays off.