Prompt engineering
Deliberately designing and checking tasks, context and output rules for a generative model.
What is prompt engineering?
Prompt engineering systematically designs and tries inputs for generative models. It includes task descriptions, required context, suitable examples and output rules. Variants are compared on appropriate cases and recorded traceably. The inputs change; this does not adjust model parameters as training does. Good wording alone does not guarantee business correctness.
In a fictional document workflow, the prompt requires document_type to be invoice, delivery_note or review. Mixed or unclear documents should go to review. The team tries the rule on clear and difficult examples. It checks both allowed output form and actual classification. A valid field containing an allowed value does not by itself establish that the document was classified correctly.
Sources
Related terms
- PromptA prompt is the input to a generative model. A clear task helps but does not replace data connections or access permissions.
- Large language modelA large language model processes language using learned patterns. Fluent text proves neither factual accuracy nor access to current business data.
- Fine-tuningFine-tuning adapts a pretrained model through further training. Supplying current documents at runtime is a different operation.
- AI hallucinationAn AI hallucination is plausible-looking content that is factually false or invented. Fluent language is not evidence of correctness.
- Data validationData validation checks data against defined requirements such as format, required fields or permitted value ranges.
- Model cardA model card documents an AI model’s purpose, foundations, evaluation conditions and known limitations for a particular version.
Articles on this topic
AI & Automation with ASES
AI assistants and automated workflows