You are here:
Search Amazon S3 Files by Meaning
Find information inside documents, audio, video, and images stored in Amazon S3 by using natural-language queries.
When your agency stores case files in Amazon S3, you can enable External Storage Search so that your team finds information by meaning, not just exact keywords. Investigators and analysts enter a natural-language query to find specific information inside any file—text in documents, spoken words in audio or video, and text in images—without opening each file individually.
Search understands intent, not just keywords. For example, a search for "fled" also surfaces files that mention "absconded" or a video where someone says "ran away." Results are ranked by relevance and include content snippets and timestamps that link directly to the source—the page in a document or the moment in a recording. You can scope a search to a single Salesforce record.
External Storage Search uses your existing Amazon S3 connection with an AWS search service. It layers full-text indexing, optical character recognition (OCR), automatic audio and video transcription, and semantic understanding on top of your stored files. The feature supports phrase and wildcard queries. You can use either Amazon Bedrock or Amazon Kendra as the search service. Users search file content by conversing with an Agentforce agent or directly from a record page using the External Storage Semantic Search component.
