From Physician to Innovator: The Next Leap in Healthcare with AI Scribes and Built-In CPT/E&M Coding
By: Steven L. Sivak, MD FACP, Sep 27, 2025
As a primary care physician with over four decades of practice, I’ve personally experienced the ever-evolving world of clinical documentation and billing. From handwritten notes to digital templates, and from ICD-9 to CPT and E&M codes, each shift brought new demands—especially in how we capture, document, and bill for patient care. That experience drove me to found Billitrite, an app designed to help physicians accurately select and apply CPT and E&M codes without the cognitive overload or confusion many clinicians face about the rules around billing and coding.
My goal with Billitrite was simple: reduce errors, ensure compliance, and lighten the cognitive burden on clinicians. But a year into that journey, I encountered a transformational innovation—ambient listening and AI-based scribing. Suddenly, it became clear that the very encounter I had worked so hard to code manually could now be transcribed, summarized, and understood by machines. And if a machine could understand the content of a visit well enough to generate a SOAP note, why not teach it to assign a billing code as well?
This realization would change everything.
Along the way, I met Walter Yuan, the president of Empathia.ai, an emerging leader in the ambient scribe space. Walter shared a vision of clinical documentation powered by AI—but also understood the critical gap between what engineers could build and what physicians actually needed. Recognizing our shared goals, he invited me to join Empathia.ai as Chief Medical Officer.
Together, we’re working to incorporate AI-generated billing code recommendations into Empathia’s ambient scribe platform. This would relieve physicians not only of documentation burdens but also from the task of determining and entering the correct billing code—tasks that are both error-prone and deeply frustrating to many of us in practice. Our vision: a hands-free, keyboard-free workflow where the encounter is naturally conducted, fully documented, and accurately billed—without ever breaking eye contact with the patient.
Technology engineers are brilliant, but many lack the deep, lived-in understanding of the physician’s day—the interruptions, the multitasking, the need to think on your feet while simultaneously clicking through templates, drop-downs, and ICD code search fields. As a practicing physician, I know that documentation is not just about checkboxes. It’s about nuance. A cough can mean reflux, post-nasal drip, or a neoplasm. Coding that visit correctly depends on what’s asked, what’s documented, and what’s inferred from the conversation.
This insight is often missed when AI tools are designed without physician involvement. Developers may see the problem as one of data capture and pattern recognition. But we see it as one of workflow, trust, and cognitive bandwidth. Without a clinician guiding the AI’s development, these tools often fall short—clunky interfaces, misaligned note structures, or misinterpreted clinical significance.
Ambient AI scribing—using microphones to capture and transcribe the patient encounter—has already shown promise in reducing documentation time. Products like those from Empathia.ai, and Nuance DAX are making headlines for their ability to generate structured notes from unstructured conversation. A 2023 study in JAMA Internal Medicine found that AI scribes could reduce documentation time by over 70%, helping physicians reclaim hours of lost time each week.
But documentation is only one piece of the burden. Medical billing adds a second layer of complexity, especially with the shift in 2021 to time-based and medical decision- making-based E&M coding. Physicians must now make real-time decisions not only about diagnoses but about how to document the complexity of care rendered to justify reimbursement. Errors are costly. Underbilling leaves revenue on the table; overbilling risks audits and penalties.
This is where our work at Empathia.ai takes the next leap: using AI not only to transcribe but to code. By pairing natural language processing (NLP) with CPT and E&M logic trees, we’re creating a system that can listen to a visit, generate a SOAP note, and recommend an accurate billing code based on documentation—all in real time.
This isn’t just about billing. It's about returning physicians to the practice of medicine. By removing the need to click, type, and toggle between screens, ambient AI and auto-coding allow for more eye contact, more empathy, and more accurate information capture. Patient satisfaction improves. Physician burnout decreases. Practices become more efficient and compliant.
It’s also a win for payers and regulators. Consistent coding driven by objective NLP models reduces fraud, improves audit transparency, and allows for better benchmarking of care quality. It can even help uncover population health trends by structuring unstructured clinical data at scale.
To build healthcare technology that works, we must bridge the gap between engineers and clinicians. When physicians are at the table—as founders, advisors, designers, or CMOs—we can align innovation with real clinical pain points. We can build tools that fit into our workflow instead of forcing us to adapt to theirs. My journey from a practicing internist to CPT educator, app developer, and now Chief Medical Officer at Empathia.ai has taught me that the best technology doesn't come from the lab. It comes from the clinic. It comes from listening—to our physician and to each other.
And with AI at our side—not in front of us—we can finally return to what drew us into medicine in the first place: the human connection.
References JAMA Internal Medicine. “Impact of AI Scribes on Physician Documentation Time.” 2023. Time Magazine. “AI Medical Scribes Are Changing How Doctors Document Care.” 2024.
