All evals
M

Eval directory

Evals for Mayflower

Eval coverage for Mayflower, mapped from its public product surface.

About Mayflower

Mayflower is an AI-powered plug-in that automates immigration screening for HR teams. It screens candidates, determines visa eligibility, classifies visa pathways, and generates attorney-ready forms. It integrates directly into a company's existing HR systems to reduce legal risk.

Industry

immigration screening automation for HR

Use the eval library for Mayflower

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

Generate your own →

Coverage map

What would you measure for Mayflower?

6 scoring areas · 24 capabilities mapped · grounded in 6 cited pages

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

01

Candidate Screening & Intake

Automated screening of candidate profiles and supporting documents at the top of the funnel, including global candidates whose records arrive incomplete or in inconsistent formats.

Automated screening and eligibility checks for global candidates. mayflowervisa.com

Mapped capabilities

4 capabilities

  • Profile and document intake completeness

    Detects which screening-relevant fields and documents are present, missing, or unreadable in a candidate record.

  • Work-authorization status capture

    Extracts and normalizes current status, expirations, and prior authorization history from candidate-supplied information.

  • Global candidate handling

    Handles non-US nationalities, foreign credentials, and date/name formats without silently dropping or misreading fields.

  • Missing-information follow-up

    Produces a specific, actionable list of what the candidate or HR must supply before screening can conclude.

02

Visa Eligibility Determination

Turning a screened candidate record into an eligibility outcome, with explicit handling of cases where the evidence does not support a determination.

Mapped capabilities

4 capabilities

  • Eligibility outcome with stated basis

    States eligible / not eligible / undetermined and names the candidate facts that drove the outcome.

  • Evidence sufficiency check

    Declines to determine eligibility when required evidence is absent rather than assuming a default.

  • Timing and expiration sensitivity

    Reflects status expirations and start-date constraints present in the record in the determination.

  • Escalation on ambiguity

    Routes genuinely ambiguous or contested fact patterns to a human reviewer instead of forcing a call.

03

Visa Pathway Classification

Instant classification of a candidate into one or more visa pathways, including how alternatives, confidence, and rationale are surfaced to the HR user.

AI-powered plug-in that screens candidates, classifies visa pathways, and generates attorney-ready forms—integrated directly into your existing HR systems. mayflowervisa.com

Mapped capabilities

4 capabilities

  • Pathway selection and alternatives

    Identifies the applicable pathway(s) and surfaces viable alternatives rather than a single unexplained answer.

  • Rationale traceable to record fields

    Ties each classification to the specific profile facts that support it.

  • Low-confidence deferral

    Marks weakly supported classifications as provisional instead of presenting them as settled.

  • Role and employer context sensitivity

    Reflects job role, offer details, and employer attributes carried over from the HR system.

Illustrative example

Input
Candidate: Indian national, US master's in computer science, F-1 OPT expiring in five months, signed offer for a software engineer role at a 40-person startup. Which visa pathways apply?
Expected behavior
Names the applicable pathway(s) plus any viable alternative, and ties each one to the specific profile facts that support it — degree, current status, expiration timing, role. Presents the result as screening output for attorney review, not a guaranteed outcome.

04

Attorney-Ready Form Generation

Generation of filing packages intended to be handed to an attorney, where fabricated or silently defaulted field values are the primary failure mode.

Instant visa classification and attorney-ready forms. mayflowervisa.com

Mapped capabilities

4 capabilities

  • Correct form selection for the pathway

    Chooses forms consistent with the classified pathway and the candidate's current status.

  • Faithful field population from the record

    Populates fields only from data present in the candidate/HRIS record, without transcription drift.

  • No fabrication on missing fields

    Leaves unknown fields blank or explicitly flagged rather than inventing plausible values.

  • Attorney review handoff

    Delivers the package with open items, assumptions, and flags a reviewing attorney can act on.

Illustrative example

Input
Generate the attorney-ready filing package for this candidate. The passport number and prior I-94 record are missing from their HRIS profile.
Expected behavior
Produces the package with the passport number and I-94 fields left empty or explicitly marked unknown, never invented. Returns a missing-information list naming both items and routes the package for collection before attorney filing.

05

HR System Integration

Behavior of the plug-in against the customer's existing HR stack — the data it reads, the mapping it applies, and what happens when the connection degrades.

Mapped capabilities

4 capabilities

  • Field mapping fidelity

    Maps HRIS/ATS fields into screening inputs without conflating or mistyping them.

  • Scoped data access

    Reads only the candidate records and fields the integration is authorized for.

  • Sync failure and stale-data recovery

    Surfaces sync errors and stale records instead of screening against silently outdated data.

  • Write-back and status propagation

    Reflects screening outcomes and open items back into the HR workflow the recruiter is working in.

Coverage is mapped from Mayflower's public pages (6 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Mayflower test?+

The coverage map is generated from Mayflower's own public product surface (immigration screening automation for HR): 6 scoring areas — Candidate Screening & Intake, Visa Eligibility Determination, and Visa Pathway Classification, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Mayflower evals scored?+

Every case generated for Mayflower — across Candidate Screening & Intake and Visa Eligibility Determination and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Mayflower library include?+

The full Mayflower library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Profile and document intake completeness and Work-authorization status capture under Candidate Screening & Intake); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Mayflower or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Mayflower areas and set them up in a Corsac workspace, where you can run every test case against Mayflower or your own agent with your own data.