Airside Labs - AI Systems for Travel & Aviation
    ChatGPT Said the Visa Would Be Accepted. It Wasn't.
    Newsletter

    Never miss an episode

    Get new Travel Tech Podcast episodes, aviation AI insights, and use-case deep dives delivered to your inbox.

    Episode 33
    September 15, 2026

    ChatGPT Said the Visa Would Be Accepted. It Wasn't.

    Rob Lee, Co-Founder & CEO, Swift Passport & Visa Services, Co-Founder & CEO, Swift Passport & Visa Services

    AI told a traveler their visa was fine. They were turned away at the border. Rob Lee, Co-Founder & CEO of Swift Passport & Visa Services, unpacks why general AI models get visa requirements dangerously wrong, how the shift from paper visas to eVisas reshuffled an entire industry, and why a database that would have cost $780,000 in 2018 is now buildable for a fraction of that, with a human still checking every single application.

    AI Visa Misinformation
    eVisa Disruption
    Business Travel Logistics
    Human in the Loop AI

    Show Notes

    Rob Lee is Co-Founder and CEO of Swift Passport & Visa Services, a company he built from a two person Chicago apartment operation in 2008, after almost being refused entry at Johannesburg immigration, into an 18 year old business serving corporate travelers, airlines, and individual applicants across dozens of countries. Rob traces Swift's early competitive edge to same day submission: while larger competitors sat on high volume mail for a day or two, Swift's small batch size let them run the full consulate circuit each morning, shaving a full day off processing times that mattered enormously to business travelers chasing last minute trips to China or Brazil. He then charts the industry shaking shift to eVisas and ETAs from around 2018 onward, which initially looked existential for passport services but introduced a new layer of complexity. India's fraud checks, for instance, can take up to 15 payment attempts before a card clears. The conversation turns sharply to AI reliability: Rob explains that general purpose AI models routinely give incorrect visa guidance because the underlying web content is outdated or contradictory, creating real consequences for travelers who act on it. Swift's own approach keeps customer data out of public AI tools entirely, while using AI heavily for internal tooling, scraping government and consulate websites to build a live requirements database that a human reviews before any application moves forward. A database Rob priced at roughly $780,000 in 2018 is now achievable at a small fraction of that cost, and the platform can prefill forms and letters of intent automatically. He closes by noting that visa processing delays are often a diplomatic reciprocity mechanism between governments, not a technical problem waiting for a software fix.

    Click to expand the full episode transcript (7,161 words approx.)

    Previous Episode

    Ep 32: The Deal AI Can Never Find

    Next Episode

    Ep 34: Trust as Infrastructure: Who Your AI Travel Agent Actually Works For

    Building AI for Aviation?

    Airside Labs helps aviation organisations validate AI systems before deployment. From adversarial testing to compliance evaluation, we bring 25+ years of aviation data expertise to ensure your AI is safe, compliant, and production-ready.

    Book A Demo

    About the Travel Tech Podcast

    Hosted by Alex Brooker, founder of Airside Labs, the Travel Tech Podcast features authentic long-form conversations with leaders, innovators, and startup founders shaping the future of aviation and travel technology. No scripts, no sponsors, no sales pitches — just real conversations about building technology that matters.