In the second article of this two-part series, we dive behind the scenes of AI Agents, examining the technology, software paradigm shifts, and architectural choices that make AI-powered automation not only possible, but essential for the future of healthcare operations.
In Part 1 of this series, we explored how AI agents are fundamentally changing the way people work across healthcare—freeing staff from repetitive tasks, improving productivity, and unlocking capacity systemwide. But how is this transformation actually happening behind the scenes? What makes these agents capable of integrating seamlessly into complex workflows, scaling effortlessly, and delivering real business outcomes? In Part 2, we’ll look under the hood—examining the technology, software paradigm shifts, and architectural choices that make AI-powered automation not only possible, but essential for the future of healthcare operations.
Did you miss Part 1? Catch up here.
Historically, enterprise healthcare software required human staff to learn and adopt new tools. This is an expensive thing to do and means that new software must only be introduced sparingly.
AI Agents change that. They don’t replace the EHR or ask staff to train to use different tools. They work inside the EHR, contributing to existing workflows, just like a human would.
That means:
• Reviewing and completing a work queue task in Epic or Cerner.
• Entering faxed referrals directly into the EHR the moment they arrive.
• Abstracting unstructured data in charts and writing back to structured fields — automatically, accurately, at scale.
These Agents aren’t asking staff to switch tools. They’re taking tasks off staff’s plates by performing the work alongside staff, in the systems and workflows that staff already use.
This is a profound shift. It means that deploying AI does not mean trying to get busy people to change their behavior. Instead it means that busy humans can have their efforts amplified by AI, within their existing workflows. And, we’re automating the tasks humans don’t want to do, or that health systems can’t afford to have humans do.
Today, as high as 35% of inbound phone calls and portal messages to health systems are status checks—patients wondering about their referrals, authorizations, or medications refills. That’s because processes are sequential and slow. Referrals sit in fax inboxes. Authorizations wait on documentation.
What if that changed?
By operating within existing systems and automating tasks in real time, AI Agents introduce more than just convenience—they introduce a new way to drive better operations and service levels. Instead of linear, manual workflow steps that rely on handoffs and delays, workflows can now run concurrently, automatically, and without friction.
For example, rather than someone waiting for weeks about the status of their referral while it sits on a fax machine, an AI Agent can be outreaching to them to schedule the moment that referral hits the fax machine. This unlocks a powerful shift: healthcare operations no longer have to move sequentially. They can happen in parallel – radically reshaping the care experience.
Not only do patients get care faster—but staff are freed up to focus on complex, high-touch care, rather than doing data entry, chasing paperwork, or filling out endless forms.
This evolution isn’t just technical—it’s economic. As Box CEO Aaron Levie has pointed out, AI agents will force a transformation in enterprise software business models.
Rather than pricing based on users or seats, the future of AI-driven software will be usage-based, output-based, and value-based. If agents can perform tasks at scale and deliver measurable business outcomes, there’s no reason to tie pricing to human headcount.
This change enables unlimited scalability. A hospital system doesn’t need to buy a license for every staff member. Instead, it can deploy as many agents as needed—aligned to the outcomes they want to achieve—and pay for results, not access.
It’s a more flexible, more aligned, and ultimately more powerful model. And it’s tailor-made for industries like healthcare, where task volume and labor constraints have historically been limiting factors.
We’ve spent nearly a decade building for this moment.
• Deep integrations into healthcare systems like EHRs to read and write data like a human.
• Specialized skills built specifically for healthcare—from prior auth submission to document transcription to chart review.
• Omnichannel patient engagement: voice, SMS, chat, and more.
• Flow Builder for orchestrating all of it with minimal lift from IT teams.
• Tools to enable effective and responsible AI, including reinforcement learning, batch testing, and dynamic LLM selection built into the platform.
• Hard-won insights from nearly a decade of at-scale, enterprise deployments at the US’s largest health systems.
This isn’t theory. This is real. And it’s already reshaping what’s possible in healthcare.