Glossary
What is Prompt Engineering?
Designing and testing the instructions and examples given to an AI model so that it produces reliable, useful results.
In plain words
A prompt is the instruction you give a language model. Prompt engineering is the work of writing, testing and improving those instructions until the results are reliably good. It is closer to writing a clear brief and then testing it than to any magic phrase.
What good prompts usually contain
- The role and goal, in plain words
- The exact format wanted for the answer
- Rules and limits, including what to do when unsure
- A few examples of good answers
- The material to work from, such as the text to summarise
The part people skip: testing
A prompt that works on three examples can fail on the fourth. Professional teams keep a set of realistic test cases and rerun them whenever a prompt or model changes, so they can see whether things got better or worse.
Security
If the model reads text from outside, such as an email or a web page, that text can contain instructions that try to take over its behaviour. This is called prompt injection. A good design does not rely on the prompt alone to keep things safe: it also limits what the system is allowed to do.
Common mistakes
- Judging a prompt by a few lucky examples
- Long, tangled instructions nobody can maintain
- Putting secrets or customer data into prompts carelessly
- Relying on the prompt as the only safety measure
Questions to ask a company that offers this
- How do you test prompts, and on how many realistic examples?
- How do you notice when a model update changes behaviour?
- How do you protect against instructions hidden in the content it reads?
- Who maintains the prompts after launch?
A good company will answer these plainly and in writing. Vague or evasive answers are worth noting.
Related terms
Find companies
Explanation last read through on 13 October 2026. It is general guidance in plain language, not advice for your particular project. Tell us if something is wrong or unclear.