PII Redaction in Azure AI Search Using Built-in Skill

Quick Answer

The built-in PII detection skill fits this requirement because it runs inside the Azure AI Search indexing pipeline itself, meaning PII gets identified and handled as documents are being processed for the index rather than needing a separate pre-processing step built and maintained outside the search pipeline. The skill recognizes common PII entity types — names, phone numbers, email addresses, and similar personal identifiers — as documents move through indexing, and its redactionMode configuration is what turns detection into actual redaction, replacing each identified entity with a placeholder before the content is stored in the searchable index. This is what satisfies the requirement to redact PII before indexing specifically, rather than after the fact: once content lands in the index, any PII within it has already been stripped out, so there's no window where sensitive raw text sits searchable and unprotected. The general pattern worth recognizing is that Azure AI Search's skillset pipeline can enrich or transform content — detecting language, extracting entities, or in this case redacting PII — as a normal part of indexing, which is why built-in skills are usually the answer whenever a requirement describes content needing modification specifically before it becomes searchable.