Conversations That Matter, a podcast from Uniphore

A handful of genuinely useful ideas emerge - the healthcare interoperability waiting game and the 'script before flows' rule for conversation designers - but they are buried under significant filler, host interjections, and the entirely substance-free rapid-fire segment that takes up a substantial portion of the runtime.

“why are you going to build your own highway when the government's got a highway coming? You just got to wait a few more years”

“script it first. Don't jump to flows, don't jump to logic. Yes, you can like sort of generally get a sense of like what are the use cases that we're going to be encompassing here. But if you don't figure out what the rhythm of the conversation is”

The government interoperability 'don't build your own highway' framing is a genuinely fresh and practical caution for healthcare AI buyers, but the rest - AI is about people, change management drives adoption, human-centered design matters - recycles standard takes without adding new angles or first-principles reasoning.

“you don't want to do too much because in five years, like, why are you going to build your own highway when the government's got a highway coming?”

“the goal is to get people to do their jobs more efficiently and honestly, uh, not spend so much time doing the stuff that kind of slows them down”

Diana Deibel is a legitimate practitioner who built real IVR/VUI products in healthcare and co-authored a respected field reference, giving her credible hands-on credentials; however, she operates at a 20-person boutique consultancy with NDA-constrained client disclosure, limiting the depth of scalable, attributable experience she can share.

“it quickly evolved into a product that was essentially following folks when they were leaving the hospital after, uh, anything that the hospital could get financially dinged for”

“we need to figure out how to streamline surgery, elective surgery, across 50 different sites”

The episode has some concrete anchors - named companies (Emmy/Wolters Kluwer), specific conditions covered, a 30-to-45-day follow-up window, and an approximate 2012 timeline - but no hard metrics, outcomes data, or named client examples, and the NDA blanket blocks most granular evidence.

“things like pneumonia, um, asthma, uh, stroke, heart attack. We created this product line that would check in with folks, um, via an IVR like system”

“rather than the nurses having to be the ones to call every single person who gets discharged with any one of these very common things for the next 30 to 45 days”

The host frequently injects personal anecdotes that stall momentum, misidentifies the guest's conference, and devotes the back third of the episode to a game-show rapid-fire with zero B2B relevance; follow-up questions exist but rarely push beyond the surface, and no claim is meaningfully challenged.

“my wife is a nurse practitioner over at, uh, Lucille Packard Stanford Children's, uh, right around the corner from our offices here”

“um, you attended a conference recently. I think it was the Voice AI conference”