UTILITYPOT / PART OF MODEPOT
Custom entity extraction
Define categories and extract reviewable mentions with exact source offsets.
Up to 120 source tokens. Each defined category returns at most three mentions; this is not exhaustive full-document NER.
Define 1–4 categories. Jev selects exact start/end tokens in the source. Missing or inconsistent spans are rejected; review the proposals before export.
Jev interpretation
AI receives the entire bounded text and your category definitions. Your inputs are not stored by this application or sent to analytics. TypeSafe receives submitted AI context.
Load an example or enter your source.
Result
What this tool does
Define up to four categories, with at most three selected mentions per category, in up to 120 source tokens. This is bounded custom entity extraction, not exhaustive multilingual NER. Jev selects source token boundaries; the application validates exact text and offsets. Missing, inconsistent or unsupported spans are rejected. Review every label before using exported data.
Privacy and uncertainty
Only explicit AI requests send bounded source context to Jev through our protected server. There is no account or application input-history database. AI classifications can be wrong; typed output is not proof of truth. Read privacy and operating limits.