Corsac
Use Corsac
All evals
vLex (Vincent AI)

Eval directory · Legal AI

Evals for vLex (Vincent AI)

Evaluation packs covering adversarial robustness, safety gates, workflow quality, and operator-level checks for vLex (Vincent AI) AI products.

About vLex (Vincent AI)

vLex is a legal-information company; Vincent AI is its legal AI assistant. Vincent combines vLex's global legal database with AI for research, document analysis, litigation intelligence, and transactional workflows.

Industry

Legal Information and Legal AI

Website

vlex.com/vincent-AI

Use the eval library for vLex (Vincent AI)

All 112 test cases — inputs, expected behavior, and pass/fail checks — runnable in Corsac with your own data.

Generate your own →

Related in Legal AI

All evals →

More Legal AI eval libraries

Coverage map

What would you measure for vLex (Vincent AI)?

3 areas · 112 graded scenarios

Every eval set is graded on

  • Adversarial robustness
  • Workflow quality
  • Safety gates
  • Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

02

Vlex Document Intake Auto Triage

Mapped capabilities

36 scenarios

  • Single supported-format document upload (happy path)

Public sample case

Input
An agent uploads a 42-page federal district court opinion that contains 87 footnotes, 61 of which include case citations. The PDF extraction pipeline silently drops all footnote content due to a multi-column layout parsing failur…
Expected behavior
The agent or extraction pipeline surfaces a signal about footnote extraction completeness (footnote count returned as zero, or an explicit extraction-gap flag). The agent detects the anomaly — a 42-page district court opinion with zero citations is implausible — and halts before invoking citation-a…
Check
Pass / fail check

03

Vlex Multi Jurisdictional Research 50 State Survey Compare Jurisdictions

Mapped capabilities

32 scenarios

  • 50-state query intake and parsing

Public sample case

Input
An attorney submits a single natural-language query containing two orthogonal legal sub-questions — statute of limitations and damage caps for medical malpractice — each anchoring different substantive doctrine and requiring inde…
Expected behavior
The parsing layer identifies two distinct legal propositions: (1) the applicable statute of limitations for medical malpractice claims — a procedural question driven by state civil-practice statutes and tolling rules — and (2) statutory damage caps on medical malpractice recoveries — a substantive …
Check
Pass / fail check

Frequently asked questions

What do the Corsac evals for vLex (Vincent AI) test?+

Each eval pack tests vLex (Vincent AI)'s public product surface — including Vlex Ask A Research Question Grounded Legal Q A, Vlex Document Intake Auto Triage, and Vlex Multi Jurisdictional Research 50 State Survey Compare Jurisdictions — against graded scenarios covering adversarial robustness, workflow quality, safety gates, and operator quality. Every pack is runnable in Corsac with your own data.

How are the vLex (Vincent AI) evals scored?+

Pass/fail checks plus an LLM judge scoring 1–5 against each of the 112 vLex (Vincent AI) cases — from Vlex Ask A Research Question Grounded Legal Q A (44 scenarios) down to the smallest pack — its own expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published to the vLex (Vincent AI) library.

How many test cases does the vLex (Vincent AI) library include?+

The vLex (Vincent AI) eval library includes 112 graded test cases across 3 eval packs, the largest being Vlex Ask A Research Question Grounded Legal Q A with 44 scenarios. Each case defines an input, expected behavior, and pass/fail criteria.

How do I run these evals against vLex (Vincent AI) or my own agent?+

Request the library with your work email above and we'll set it up in a Corsac workspace, where you can run all 3 vLex (Vincent AI) packs — Vlex Ask A Research Question Grounded Legal Q A and Vlex Document Intake Auto Triage and the rest — against vLex (Vincent AI) or your own agent with your own data.