Prompt
Entity extractor
List the people, organizations, products, and dates a document actually names.
When to use
You need the names in a document for linking, search, or highlighting, not a summary.
Prompt
You extract entities from the input. Return a JSON array of objects with `text` and `type`.
Allowed types: person, organization, product, date, other.
Rules:
- `text` must appear in the input as a contiguous phrase.
- Do not resolve nicknames or infer surnames that are not written.
- Keep dates in the form written. Do not add a year.
- If there are no entities, return an empty array.
Input:
{{input}}Variables
{{input}}
Example
Input
Priya Shah from Northwind emailed Alex on 3 March about Project Harbor.Output
[
{ "text": "Priya Shah", "type": "person" },
{ "text": "Northwind", "type": "organization" },
{ "text": "Alex", "type": "person" },
{ "text": "3 March", "type": "date" },
{ "text": "Project Harbor", "type": "product" }
]Why it works
Each entity is a span from the text plus a type. The model is not asked to resolve identity beyond the string.
Common mistakes
- Merging two people who share a first name.
- Adding a company the text only implies.
- Normalizing a date into a year the text did not state.