I Can't Code. I Built a SaaS. Then the Security Questions Started.
What happens after AI helps you build and ship a product, when a real enterprise customer starts asking the hard questions?
The speaker is a safety consultant, not a developer. Using AI, he built and launched a commercial risk-diagramming SaaS that a critical-infrastructure operator asked to buy.
Building it turned out to be the easy part.
Then came the formal security assessment: a questionnaire full of questions he could barely answer, and a review that uncovered four real defects. None of them unusual. Every one the kind of thing an AI-assisted build carries quietly, unnoticed, until somebody looks. All four have since been fixed and evidenced.
In this highly practical session, he'll share what enterprise buyers actually look for before approving a SaaS product, the real issues uncovered in an AI-assisted Next.js, Supabase, Vercel and Stripe stack, and how AI helped not only to build the product, but also to diagnose and fix it.
And it won't just be slides.
Two tests will be run live on the real product: first, reading out where the application code is actually executing; then, an attempt to break its data isolation. A third, recorded beforehand, shows the tests designed to catch that failure being deliberately broken, demonstrating why simply having tests is not the same as knowing they actually work.
Designed for founders, makers and AI-assisted builders who can ship fast, but haven't yet had an enterprise security or procurement team look under the hood.
What you'll take away: a clearer understanding of the questions enterprise buyers ask before they buy, common weaknesses in AI-built SaaS products, four real defects you can check in your own stack immediately, and a practical method for proving that your tests can actually detect failure.
Presenter: Rod Cameron