Research & assessment

How Visa Engine researches immigration rules

This page explains how Visa Engine produces its guides and assessments. It is the difference between a regulatory requirement, agency guidance, precedent, common evidence practice, and a Visa Engine score.

What Visa Engine is — and is not

Visa Engine is software for preparing talent petitions, skilled-work applications, and visitor checklists. Analysis is checked against published immigration rules stored in our knowledge base — not a generic chatbot’s memory.

Visa Engine is not a law firm, not legal representation, and not a substitute for advice from a qualified immigration attorney. We do not file petitions. Your lawyer still files.

How we separate sources

Immigration law is not a single rules engine. We try to keep these layers distinct:

Regulatory requirement

The controlling instrument — for example 8 CFR, the Immigration Rules, the Visa Code, IRPA, or the Migration Act. This is what the law actually requires.

Agency guidance

How the agency says officers should apply that instrument — USCIS Policy Manual, GOV.UK guidance, IRCC program pages, Home Affairs instructions.

Precedent

Decisions that shape adjudication where they apply — for example Matter of Dhanasar for NIW, or the Kazarian two-step analysis for extraordinary ability. We name precedent only when the guide already relies on it.

Common evidence practice

How petitions are typically documented in practice: what tends to be treated as strong versus weak evidence. This is not a legal requirement.

Visa Engine assessment

Our product’s score, gap list, and RFE-risk hints. These are heuristics grounded in the sources above. They are not a USCIS, Home Office, IRCC, or Home Affairs decision.

How guides are researched

For each visa route we identify the controlling regulation, check current agency guidance, and note relevant precedent where it actually governs the analysis. We separate what the law requires from what is commonly used as evidence.

Guide pages show a primary-authority box with those sources. We revise content when major policy changes land. Dates on individual pages are when that page’s prose was last edited — not a claim that every citation was re-verified that day.

How AI analysis is grounded

When you upload a document, the model is steered with retrieved excerpts from our curated visa knowledge — policy, criterion, RFE, and decision notes tied to that visa type. It is not asked to recall immigration law from training data alone. Retrieval can still miss a source or over-weight a chunk. Treat the output as a structured second read, not an official determination.

How to read scores and RFE risk

Readiness scores, criterion coverage, and RFE-risk hints estimate how complete and coherent a file looks against published criteria or a visitor checklist. They are not approval probabilities, processing-time estimates, or predictions of what an officer will do. Weak or missing evidence raises assessed risk; strong evidence does not guarantee approval.

Limitations

Rules change. Member-state and consular practice varies. Our knowledge base is curated prose, not a complete tree of every CFR subsection or AAO decision. Human review — by you and, where appropriate, counsel — remains the last step before filing.

Last reviewed: August 16, 2026.

Educational only — not legal advice.