JevMade Sign in
← Back to experiments

Apps & data pipelines

RazorRisk

Adds an optional second opinion to fraud investigations, comparing Jev's action and grounding check with the existing evidence pipeline.

Bookmark: RazorRisk Keep this in your collection.
Leave a noteWhat would you try with this? : RazorRisk

Only you can see your notes.

Source screenshot of RazorRisk
SOURCE SCREENSHOTFull screenshot ↗

What it does

Jev never sees the investigator's answer and does not change the fraud score. Disagreement sends the case to review, while failures leave the original report intact.

How you can use it

Before touching any code, outline the specific fraud signs your team reviews on suspicious transactions. Have a developer supply an access key from TypeSafe and turn on the verification setting. The prototype then analyzes transaction records and drafts an explanation for human review.

This extra step acts as an optional audit pass. It checks whether the drafted explanation actually matches the gathered payment evidence. It never changes the final fraud risk score or blocks reports if the check fails, leaving final decisions to your investigators.

Maker-reported (not independently measured by JevMade): Investigation latency without Jev: median 38.60 ms, p95 136.85 ms over 3 runs, measured locally on 2026-09-20 (author-reported) · With Jev: not measured — the README states no provider-backed run has been claimed yet · Auto-resolution requires Jev's independent action confidence to be at least 0.85 and no mandatory-human reason to be present

Primitives
choice, noul
Platform
Python
Added
Project created
GitHub stars
0 (snapshot, not live)

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.