Skip to main content

AI glossary Models & Architecture

What is a prompt in AI?

← All glossary terms
In short

The input given to an AI model, such as instructions, context, examples and the question, that guides what it produces.

Explained

A prompt is everything a generative AI model receives as input before it responds: the instructions, the context, any examples and the user's request. It can be text alone or include images, documents and other files. The model has no other way of knowing what is wanted, so the content and structure of the prompt strongly shape the quality and consistency of the answer. In applications, the prompt is usually assembled by software from several parts, such as a fixed system prompt, retrieved documents and the user's message, rather than typed by hand each time.

What a good prompt contains

  • Role and goal: what the model is for in this task and what a good result looks like.
  • Context: the relevant facts, documents or data the answer should be based on.
  • Instructions and constraints: steps to follow, what to avoid, tone and length.
  • Examples: one or a few input and output pairs that show the expected format.
  • Output format: for instance a list, a table or JSON with specific fields.

System prompt vs user prompt

The system prompt is set by whoever builds the application. It defines the model's role, rules and boundaries and stays the same across conversations. The user prompt is what the person types or what the application passes in for each request.

In production the final prompt often combines both with retrieved content and tool results. Because outside content ends up inside the prompt, it can also carry hidden instructions, which is the root of prompt injection.

Prompts in production

  • Treat prompts as code: version them, review changes and keep them in a repository.
  • Test changes against an evaluation set before releasing them.
  • Keep the stable parts first and the variable parts last, which helps caching and readability.
  • Watch length: more context is not always better, and long prompts cost more and can dilute the key instruction.

Frequently asked questions

What is an example of a prompt?

'You are a customer service assistant. Using only the policy below, answer the customer's question in two or three sentences. If the policy does not cover it, say so. Policy: [...] Question: [...]' It sets a role, context, a constraint and a format.

What is the difference between a prompt and prompt engineering?

A prompt is the input itself. Prompt engineering is the practice of designing, testing and refining prompts so a model gives reliable results for a task.

What is a system prompt?

A system prompt is a set of instructions defined by the application, not the end user, that tells the model how to behave throughout a conversation: its role, its rules, its tone and what it must not do.

How long should a prompt be?

As long as needed to give the model the instructions and context it cannot guess, and no longer. Clear structure matters more than length; irrelevant material can make results worse and raises cost.

Can a prompt include images or documents?

Yes. Multimodal models accept images, PDFs, audio and other files as part of the prompt, alongside text instructions about what to do with them.

Decision guides that use this term

Where it shows up at Thinkia

Further reading

Sources

Updated: 2026-10-21

Want to see how this applies to your organisation?

Talk to an AI Expert