Individual retriever filters and returned fields
Key Takeaway
Filters in an individual retriever are available only when the selected search index has filter fields defined, so define a field as a filter field on the search index before you filter on it.
An individual retriever takes a general-purpose Data 360 search index and narrows it to one specific use case. A prompt template, agent, or flow then gets only the content it needs for grounding. You build an individual retriever in AI Models (formerly Einstein Studio). Filters narrow which records can come back. Returned fields set what information reaches the prompt and how many results the prompt gets.
What an individual retriever does
A search index is often built for broad use, such as indexing an entire knowledge base. An individual retriever narrows that index for one job. For example, an individual retriever can return:
- Only one category of articles
- Only the most recent content, such as the last 90 days
- Only content for one geographic region
To make returned content more relevant for a particular use case, create a custom individual retriever.