What is a prompt?
A prompt is the instructions, context and examples given to an AI model to produce a specific output. It can be a one-line question or a long brief with background, data, the desired format and examples of good answers. OpenAI defines prompt engineering as the process of writing effective instructions for a model, so that its output consistently meets your requirements.
Why it matters to a revenue leader
The quality of AI output tracks the quality of the brief, the same as with a new hire. Vague prompts get generic answers; specific prompts with the right data get useful ones. Prompts are also an asset and a risk: a well-built prompt captures how your team thinks about a deal or an account, and that thinking travels with it when it is shared. Anthropic's documentation advises defining success criteria, and a way to test against them, before refining a prompt.
Example
Two managers ask an AI tool to review the same service agreement renewal. The first types "Is this renewal at risk?" and gets a generic list of risk factors. The second includes the service history, open deficiencies, last year's price increase and the customer's recent emails, and asks for the three biggest risks with evidence for each. Only the second answer is specific enough to act on.
"Where do the prompts behind our AI tools live, who wrote them and how do we know they still work?"
Sources
- Prompt engineering | OpenAI API checked 9 Oct 2026
- Prompt engineering overview | Claude Platform Docs checked 9 Oct 2026