SeamlessVisa Form Engine

Project Overview

SeamlessVisa is a technology-powered global mobility platform covering immigration, visa processing, study abroad, and interview preparation. The Form Engine is one of its core products — a guided, reactive form experience for visa and immigration applications.

I was part of the front-end engineering team at SeamlessVisa, focused on the form's behaviour: how it screens users before they begin, how it adapts to their answers as they go, how it catches errors and contradictions, and how it hands off to the AI review at the end.

The form started as a US-specific product and has since expanded to cover Canada, Australia, and additional destinations as the platform's service offering grew.

How the Form Works

• Eligibility quiz — before the full form, applicants answer a set of screening questions. Any answer that signals a disqualifying condition ends the flow early with an explanation, rather than letting the applicant fill out the entire form only to be rejected at submission
• Conditional visibility — as the applicant answers questions, irrelevant sections disappear automatically. If they indicate they're single, all marriage-related questions are hidden. If they've never had children, all children-related questions are hidden. Applicants only ever see what actually applies to them
• Cross-field contradiction detection — the form watches for answers that conflict with each other. A name entered in one field that doesn't match a correlated field, or a date that contradicts another date, triggers a warning or an error depending on severity
• Eligibility contradiction checks — the form also checks answers against the eligibility quiz responses. If something in the main form contradicts what the applicant said during screening, they're flagged and prompted to review
• Advisor prompts — in cases where contradictions or flags reach a threshold of complexity, the form prompts the applicant to contact an advisor rather than continue alone
• AI review — on completion, every answer is passed through a FiBi-powered analysis that gives the applicant a read of their application: what to expect, where the weak points are, and what the overall picture looks like before submission

The Problem

Immigration and visa forms are genuinely complex. They are long, filled with conditional sections, and written in language that assumes familiarity with legal and immigration concepts. Applicants routinely encounter three categories of failure:

• Wasted effort on ineligible applications — without upfront screening, an applicant can spend significant time on a detailed form only to discover at the end that a single disqualifying condition makes their application non-viable. That time is completely lost.
• Confusion from irrelevant questions — a static form that shows every field regardless of context is overwhelming. Applicants who are single, childless, or have no prior travel history are confronted with entire sections that simply don't apply to them, with no guidance on what to skip and what to fill in.
• Undetected errors and contradictions — forms that don't validate across fields let applicants submit applications with internally inconsistent answers. A name discrepancy, a date that doesn't line up, an answer that contradicts something said earlier — these are the kinds of errors that cause rejections and delays that a smarter form could have caught before submission.

Standard form validation catches missing fields and format errors. It doesn't catch logic problems, contradictions, or eligibility issues. That's the gap this form engine was built to close.

Eligibility First

The form opens with a targeted eligibility quiz — a short set of questions designed to surface disqualifying conditions before the applicant invests time in the full application. Each answer is evaluated against known disqualifiers for the relevant visa or immigration category. If a disqualifying condition is detected, the flow stops, the reason is explained, and the applicant is directed to appropriate next steps rather than being left to fill out the rest of the form anyway.

This approach serves both the applicant (no wasted effort) and the platform (no ineligible applications reaching the review stage).

Reactive Conditional Visibility

The form is fully reactive. Every question that depends on a prior answer is gated — it only appears when the condition that triggers it has been met. This is applied broadly:

• A single applicant never sees marriage or spouse questions
• An applicant with no children never sees questions about dependants
• Sections conditioned on travel history, employment history, or previous visa applications only appear when the relevant prior answer indicates they apply

The result is that applicants only ever engage with the questions that are actually relevant to their situation. The form feels shorter and more personal than a static equivalent would — because it is.

Contradiction Detection

As the applicant fills in the form, answers are continuously checked against each other. Two categories of conflict are handled:

• Cross-field contradictions — answers that reference the same information in different parts of the form (names, dates, identifying details) are compared. A mismatch triggers a warning and asks the applicant to reconcile the discrepancy before continuing
• Eligibility contradictions — if a main-form answer conflicts with something stated during the eligibility quiz, the form flags it. Eligibility answers are treated as a baseline that the rest of the form must remain consistent with

Severity determines the response: warnings allow the applicant to continue after reviewing; errors block progression and must be resolved; some flags recommend contacting an advisor rather than proceeding independently.

AI Application Review

When the applicant completes the form, every answer is passed to FiBi — the platform's in-house AI — for analysis. The review reads the application holistically and gives the applicant a clear picture before they submit: what the application looks like, where the potential weak points are, and what to expect from the process. It is not a guarantee of outcome — it's a pre-submission read that helps applicants go in with realistic expectations and address anything addressable before it reaches a decision-maker.

Outcomes

• The form engine is live as part of the SeamlessVisa platform at seamlessvisa.com.
• The eligibility-first approach means disqualified applicants are identified before they invest significant time in the form — a better experience than discovering ineligibility after completing the full application.
• Conditional visibility substantially reduces the apparent complexity of the form. Applicants see a form sized to their situation rather than the full universe of possible questions.
• Contradiction detection surfaces errors that standard field-level validation misses — specifically the cross-field and eligibility-consistency errors that tend to cause downstream problems in the actual application.
• The platform has expanded beyond its original US focus to cover Canada, Australia, and additional destinations, with the form engine adapting to each jurisdiction's requirements.

Key Learnings

• Eligibility screening changes the entire user experience: Moving the disqualification check to the beginning rather than the end is not just a UI decision — it's a respect-for-the-user decision. Failing someone after they've completed a long form is a much worse experience than stopping them early with an explanation.
• Conditional form logic compounds quickly: Each new question that depends on a prior answer adds to an increasingly branching decision tree. Keeping that logic consistent — especially as new countries and visa categories are added — requires careful structure at the data and component level.
• Contradiction detection needs a severity model: Not all conflicts deserve the same response. A soft warning that lets the applicant continue is appropriate for ambiguous discrepancies; a hard block is appropriate when the contradiction indicates a genuine data integrity problem. Getting that calibration right required iteration with the team.
• Scaling across jurisdictions requires modular form design: When the form expanded from US-only to multi-country, the sections that were jurisdiction-specific needed to slot in and out cleanly without breaking the validation logic for other sections. That flexibility has to be designed in from the start.