On September 30, Paubox hosted AI + HIPAA: Healthcare's new builders, a one-hour panel on what changes when people outside engineering start building with AI.

IT staff, compliance officers, and clinicians now build intake forms, automations, and tools patients touch, often without the review steps an engineering team runs before launch.

Dawn Halpin hosted, Dean Levitt took audience questions, and the panel was:

  • Atiya Hasan, MD, MBA, MSHLS, CEO of Samar Health
  • Arvita Tripati, MBA, Founder and Principal of Vahana Labs
  • Jim St. Clair, Mississippi Department of Health and Mississippi Coding Academies
  • October Boyles, DNP, MSN, BSN, RN, ICANotes

Jim St. Clair pointed out that an AI vendor's business associate agreement (BAA) doesn't come with the documentation a typical vendor gives you on where your data goes, because a model doesn't show you that. October said BAAs are "not magic permission slips" and described clinicians pasting PHI into free AI tools as "a compliance incident wearing lip gloss."

Our takeaways

Each panelist closed with one piece of advice for anyone building with AI:

  • Start with the pain point. Watch why people paste work into ChatGPT or record notes with Siri, then build the fix around what they're trying to do. (Arvita Tripati)
  • Understand your data, and what you want to do with it, before you pick a tool. (Jim St. Clair)
  • Learn how an expert would solve the problem by hand before you automate it. (Atiya Hasan)
  • Healthcare runs on trust, so the builders who win will be the most clinically responsible ones. (October Boyles)

The recording is available on demand.

See related: The 2026 Benchmark for Building with AI in Healthcare