Why Patient Support Needs Scripts AND Conversations

The case for combining structured guidance with unstructured AI chat—because not everything should be a dialogue

By C. Anthony Jones

Ever since HAL in 2001: A Space Odyssey or Jarvis in Iron Man, we’ve been fascinated by the idea of conversational machines—AI that talks like a person and understands us effortlessly. And now, with real-world advances in AI agents, that science fiction is quickly becoming reality.

But in the rush to embrace conversational AI—either with basic chat or speech—it’s easy to forget that not every patient interaction needs to be, or should be, a conversation.

Some questions have clear answers. Some steps just need to be followed. In those cases, structured guidance is faster, simpler, and less prone to misinterpretation. The future of patient support isn’t fully automated or fully scripted. It’s both.

The Temptation of the “Uber” AI Agent

You can’t scroll a feed or skim a headline without running into hype about AI. In healthcare especially, the buzz feels breathless—an infatuation with the idea that a single, omnipotent AI agent will soon manage everything from diagnosis to discharge and beyond.

It’s easy to see the appeal. Clinical encounters—appointments, phone calls, consults—are often conversations. They’re about translating symptoms into a diagnosis and then mapping that to a treatment. This is where AI agents can (and will) do remarkable things.

But just because some clinical interactions are conversational doesn’t mean they all should be. Once a diagnosis is made or a procedure performed, most follow-up activities are highly structured. Forcing that structure into a conversational template doesn’t enhance the experience—it slows it down.

When Structure Works Best

Surrounding those clinical conversations is a long tail of action-oriented care:

• Time-sensitive prep instructions
• Medication reminders
• Fasting rules
• Physical therapy protocols

These are rule-based, time-bound, and specific. Patients don’t need a dialogue—they need clarity, timing, and confidence.

Imagine if every time you bought something that needed assembly, the instructions were delivered through a chatbot—no diagrams, no bullet points, just a meandering conversation. It wouldn’t be helpful. It would be maddening.

There’s a reason we prefer diagrams over dialogue for instructions:

• We read nearly twice as fast as we speak or listen(1)
• We process visual information in as little as 13 milliseconds(2)
• And we retain more when presented with structured, scannable content rather than a linear verbal stream(3)

For step-based medical tasks, structure isn’t just clearer—it’s cognitively lighter.

Where Conversations Are Essential

But not everything can be reduced to a checklist.

Healthcare is full of emotional ambiguity—patients who are confused, anxious, hesitant, or overwhelmed. Someone worried their pain is abnormal. Someone unsure if they followed instructions correctly. Someone afraid to admit they didn’t understand.

That’s where unstructured, conversational AI can shine.

It provides space for patients to express what’s bothering them, ask sensitive questions, or work through internal resistance. It uncovers barriers. It offers reassurance. It helps people keep going when structure alone isn’t enough.

The Risk of Over-Reliance on Either Approach

• Structured-only = efficient, but cold and brittle; difficult to capture and/or deliver important nuances.
• Chat-only = empathetic, but slow and sometimes confusing; inefficient for some communication; forces a slower, more methodical pace.
• Neither is sufficient on its own for complex, longitudinal care

Real support means knowing when to guide and when to listen.

The Hybrid Advantage

At Frontive, we’ve built our platform around this dual model:

• Structured, protocol-based guidance for delivering clear, timely, and actionable steps
• Unstructured, AI-based conversations for when patients need context, reassurance, or help making sense of what’s next

This hybrid approach offers scalability without sacrificing empathy. It lets each type of content be delivered in the form that works best—whether that’s a checklist or a conversation.

More importantly, it reflects how real people actually navigate complex care: Step-by-step when they’re confident. Back-and-forth when they’re not.

AI Should Fit the Patient—Not the Other Way Around
The goal isn’t to replace everything with conversational AI. The goal is to use it intentionally—to extend what works, fill in the gaps, and adapt to how patients actually think and behave.

The future of patient support doesn’t belong to conversational AI alone.
It belongs to clarity, empathy, and thoughtful design.

Sources:

1. Sticht, T. G., & James, J. H. (1984). Listening and Reading: Comprehension of Oral and Written Language.
2. Potter, M. C., Wyble, B., Hagmann, C. E., & McCourt, E. S. (2014). Detecting meaning in RSVP at 13 ms per picture. Attention, Perception, & Psychophysics, 76(2), 270–279.
3. Mayer, R. E., & Moreno, R. (2003). Nine ways to reduce cognitive load in multimedia learning. Educational Psychologist, 38(1), 43–52.