Writing Your First AI Prompt

A prompt is the input you provide to an AI language model: the text, question, or instruction the model uses as the basis for its response. Everything the model produces is a direct consequence of what it receives in that prompt. Understanding this requires one technical fact: large language models such as Claude, GPT, and Gemini have no inherent memory between separate conversations. Some chat tools now offer an optional memory feature that carries a few details across sessions, but by default each new session begins with a blank slate, and the model sees only what you have written in the current exchange. This is the context window, the totality of what the model can see at any one moment, and if you have not told it who you are, what you are trying to achieve, or who the audience is, then as far as the model is concerned those things do not exist. Think of the model as an exceptionally talented freelancer who arrived at your desk this morning. They are knowledgeable, well-read, and capable, but they know nothing about you, your organisation, your project, or your preferences. They have never met your clients and do not know your tone of voice. You would not hand this person a sticky note saying write something about the project and expect a polished result; you would give them a proper brief, covering who the audience is, what the goal is, what format the output should take, and what to avoid. The model is that freelancer, so brief it accordingly. There is also an important implication of how LLMs generate text: they are probabilistic systems, which means the same prompt can yield different outputs across separate runs. There is no single correct answer waiting to be unlocked by the right incantation. Prompting is a skill of clear thinking and clear instruction, not a search for a magic formula.