On Wednesday, October 28, we hosted The Paubox Table at Urban Hill in Salt Lake City, Utah.

Mountan Mitchell and Ange Beckum hosted for Paubox. Nine guests joined them from across healthcare, from IT and engineering teams to counseling, therapy, and health tech.

The mood in the room was a mix of enthusiasm and caution. Everyone is excited about what AI can already do, and just as focused on doing it safely.

What we heard

  • Nearly every conversation came back to one question: how to get the benefit of AI without losing control of it. Guests worry about staff going around approved tools, the "ghost AI" problem (read: Shadow AI), and small IT teams are leaning on approved-tool lists to keep AI use in view.
  • Guests keep a human in the loop. Trust in these tools is still being built, so they double-check AI output before acting on anything consequential.
  • Teams are automating their own work, from CRM and telephony integrations to retention dashboards and revenue-cycle tooling. If your staff is building its own tools, The CISO's guide to healthcare AI governance covers how to keep track of them.
  • Several guests saw the Paubox MCP server as a natural fit for that work. It lets Claude and other AI tools send secure, HIPAA compliant email without giving up privacy or control.
  • Privacy goes well past a compliance checkbox. Two health tech founders described how they anonymize images, video, and other media. Most people never think of those as identifiable information.
  • Therapists in guests' networks are getting real value from AI-assisted meeting recording and transcription. It helps with documentation and surfaces diagnostic and follow-up cues.
  • Legal teams' AI spend is growing fast, and several guests aren't sure the return justifies the cost.

Read more: The Paubox Table in Charlotte