Notra
Use narrow Jev judgments for brand sentiment and list position inside a broader AI visibility platform.
What Jev decides
Positive, neutral, or negative sentiment and list position from one through ten or none.
Notra first detects configured brand names and aliases in AI answers, then asks Jev for sentiment and list-position Choices while a separate general model supplies competitors, excerpts, and fallback fields.
- Jev role
- Judge how a detected brand is described and where it appears in one AI answer.
- Input
- The company name, configured aliases, buyer prompt, and full assistant answer after deterministic mention detection.
- Output
- Two independent Choice results for sentiment and list position.
- Decision frequency
- Once for each mentioned tracked brand in a completed AI answer.
- Fallback
- Skip Jev when no exact brand or alias is present; keep the general judge result when Jev is disabled, unavailable, or fails.
How it works
Overview
Notra is an open-source generative engine optimization platform with a commercial hosted product. Its wider workflow scans buyer questions across AI engines, stores answers and citations, tracks visibility, and supports content planning. Jev has a narrower role inside that system: after deterministic code finds a complete brand name or configured alias, it classifies the surrounding answer as positive, neutral, or negative and selects list position one through ten or none.
A separate general judge remains responsible for competitor extraction, excerpts, and fallback fields. If the Jev feature is disabled, credentials are unavailable, or evaluation fails, those general-judge values remain. At review time successful low-confidence Jev results could still override in-range fields, so consumers should add and evaluate their own uncertainty policy rather than treating every classification as equally reliable.
- Language
- TypeScript
- License
- AGPL-3.0
- Author
- Notra
Sources
Related projects
typesafe-computer-use
A native macOS computer-use loop built from local OCR, accessibility data, and bounded Jev actions.
What Jev decides
Which computer action, visible item, site, or off-screen accessibility control to use next.
Jev Chat Assistant
An Android conversation copilot that uses Jev to interpret chats and rank reply candidates without sending for the user.
What Jev decides
The likely intent, conversation risk, reply need, best next action, unresolved tension, and which of three candidate replies fits best.
pg-jev
Use Jev decisions directly inside PostgreSQL filters, rankings, and classifications.
What Jev decides
Whether a row matches, how strongly it matches, or which supplied label best describes it.