Searching Your Library
The examples below reflect the search behavior implemented in current private development builds.
Start with the smallest useful clue: a few words, a quoted phrase, a tag, or a file type. Memni combines full-text matches with on-device meaning-based results and ranks them together.
Useful first searches
Section titled “Useful first searches”water damage deductible"contract renewal"tag:insurance type:pdfA space between ordinary terms means AND, so water damage asks for results containing both terms. Meaning-based matches can still bring up related passages that use different wording.
Text matching
Section titled “Text matching”"water damage"— exact phrasedamag*— word prefixjewlry~— approximate spellingwater AND (damage OR deductible) NOT draft— Boolean expressiontermination NEAR/10 notice— words within a ten-term window
Filters
Section titled “Filters”Limit a term or value to a document field:
title:,body:, andmetadata:tag:,group:,type:, andsource:path:andfile:
Examples:
title:"annual report" type:pdftag:client renewalmetadata:matter=case-alphaDates and custom metadata
Section titled “Dates and custom metadata”created:2026-05-09andmodified:2026-05-11match a calendar day.metadata:policyclaimsearches text across custom metadata.metadata:matter=case-alphamatches an exact metadata key and value.
Advanced matching
Section titled “Advanced matching”Memni also supports case-sensitive terms such as case:NASA and regular expressions such as regex:"INV-[0-9]{4}". Regular expressions run against a bounded candidate set to keep searches responsive.