What is field validation?
Field validation is the narrowest form of data validation, a rule that examines one field on its own. Typical rules, often declared in the schema, confirm that a required field is filled in, that the value has the right type and length, that it matches a pattern such as a five-digit ZIP code, that it falls within a sensible range, and that it comes from an approved list such as state codes. Some identifiers carry a check digit, calculated from the other digits, which lets a rule catch most typing mistakes without looking anything up. Field validation cannot tell whether a well-formed value is the right one, so a valid but wrong account number still needs record-level or cross-system checks.
Field validation in practice
A specialty clinic receives referrals through a web form and by fax, and an AI extraction step reads the faxed pages. Field rules check that the patient’s date of birth is a real date in the past, that the member ID matches the payer’s format, and that the referring provider’s National Provider Identifier (NPI) has ten digits and a valid check digit. A value that fails is highlighted for the intake coordinator instead of being saved, so a misread digit is corrected before it turns into a denied claim. The same rules belong at the boundary of every tool an AI agent can call, where a clear rejection message lets the agent fix the value and try again.