Why Healthcare Couldn't Automate Until Now

Healthcare interactions are rarely linear. A patient calls to schedule an appointment and then asks about a current medication or mentions a new symptom. They might be anxious or unsure of what they need. All of this requires context from their EMR or CRM and the ability to adjust in real time.

Traditional IVR systems and rule-based bots weren't built for that. They can automate simple, deterministic tasks, but they break when a patient changes direction or asks something unexpected. That's why most complex workflows, including appointment scheduling and care navigation, have remained almost entirely dependent on human agents.

Modern agentic AI systems can manage unstructured, multiturn conversations, adapt when patients shift topics and coordinate work across multiple specialized AI agents. The technology has caught up, but the operational side of most health systems hasn't.

What The Transition Actually Requires

Adopting a hybrid workforce begins as an operational redesign, well before any AI touches a patient interaction.

The first step is understanding how work moves through your contact center today. Most organizations don't have a clear picture of which interactions are routine, which are complex and where handoffs between teams already break down. You can't distribute work between humans and AI if you haven't mapped the workflows you're trying to change.

From there, leaders need to define the human-AI boundary. Not every interaction should be automated, and not every interaction needs a human. Where does AI handle things autonomously? When does it escalate? What does the handoff look like? These questions require clinical, operational and compliance leadership in the same room because the answers affect patient safety, staff roles and regulatory exposure.

The hybrid workforce spans departments that don't always coordinate well. Clinical teams need to trust what AI is doing in patient-facing interactions. Operations needs to rethink staffing models. IT needs to ensure data flows between systems that have historically run in parallel. These conversations need to happen before deployment, not after.

Finally, organizations need to prepare their people. When AI absorbs routine volume, human agents take on more nuanced, emotionally sensitive work. That's a different skill set, and it requires investment in training, not just reassignment.

Where Organizations Get Stuck

The most common mistake health systems make is automating in isolated pockets. An organization deploys a bot for appointment reminders or password resets and calls it transformation, but a single-task bot is just another silo. The hybrid model only works when AI and human agents operate within the same system, sharing context and data across interactions. Otherwise, patients still repeat themselves, and staff still lacks the full picture.

Data fragmentation makes this worse. AI needs access to clinical, administrative and operational data to act with context. Many health systems still have their EMR, CRM and workforce tools running separately with limited integration. Until data flows across those boundaries, AI agents are operating without the information they need to be useful.

Then there's the governance gap. Healthcare is heavily regulated, and AI interacting with patients needs clear guardrails: what it can and can't say, how it handles sensitive information and when it must defer to a human. Organizations that skip this step put both compliance and patient trust on the line.

How It Works When Organizations Are Ready

When the internal work is done, the hybrid model produces results that neither humans nor AI can achieve alone.

One health system we worked with shared an example of an outreach call about colorectal cancer screening. The AI agent began the conversation with appropriate context, but the patient immediately shifted topics: "I'm having vision problems, and that is what I need help with." A scripted bot would've failed at that moment, as the question didn't match the script.

Because this organization had done the preparation—mapped its workflows, connected its systems, established clear escalation paths—the conversation continued. The AI agent recognized the change in intent, pulled relevant history from connected systems and gathered enough detail to understand the concern. When clinical judgment was needed, a human agent joined the interaction with full context. The patient didn't repeat a word.

That kind of experience traces back to decisions made months before the AI ever picked up the phone. Even the AI's voice mattered. Many health systems now create AI voices that reflect local dialects, building familiarity in situations that can already feel stressful. That's a deliberate organizational choice about how patients should experience the system.

What Comes Next

The workforce challenge in healthcare won't resolve itself through hiring alone. Demand will keep rising, and staffing constraints will follow. A hybrid workforce that pairs AI capacity with human judgment offers a way forward, but only if health systems are willing to do the operational work first: redesigning workflows, aligning leadership across departments, retraining staff and building governance structures that hold up under regulatory scrutiny.

Patient volume will eventually outgrow every traditional staffing model. The health systems doing this work now will already have a model that absorbs it.​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?