For years, the conversation around burnout has centered on physicians, and more recently, ambient artificial intelligence (AI) has been positioned as a solution to reduce after-hours charting and reclaim time from the electronic health record (EHR). But as these tools gain traction, a gap is becoming harder to ignore: Many are designed for a single user, even as care is delivered across teams and workflows that extend well beyond the visit. As a primary care physician, I see every day how even routine care extends beyond the visit. Referring a patient to a specialist sets off a chain of coordination, from prior authorizations to patient outreach to aligning with specialist offices on next steps. Each interaction carries documentation with it, and that work rarely sits with just one person.
This is where ambient AI has an opportunity to evolve. Not only to capture the conversation in the exam room, but to reflect the broader reality of care, including the coordination and decisions that happen between visits. Today, much of that responsibility still falls on nurses, medical assistants, and care managers who have been largely left out of the first wave of innovation. Improving care team well-being requires more than optimizing a single workflow; it depends on designing for how care is actually delivered across the entire care team.
Documentation lag creates system-wide friction.
A clinical encounter generates work far beyond the physician’s note. I recently had a complex primary care visit for a patient with diabetes and depression. During the visit, decisions were made in real time, including initiating a glucagon-like peptide-1 (GLP-1) medication, adjusting antidepressant therapy, addressing adherence barriers, and coordinating behavioral health follow-up. Each decision immediately created downstream work across the team. Documentation still lags behind decision-making: Plans are made in the visit, but recorded later, which delays action across roles. In that gap, physicians carry incomplete tasks in their heads, nurses chase clarifications, and coordinators queue work that could already be underway. What seems like a small delay quickly compounds into system-wide friction, accumulating as cognitive load across the care team.
When documentation is delayed, the system slows with it. This gap between decision and documentation is where the real opportunity for AI begins.
Clinical AI must support the entire care team.
The most impactful opportunity I see for AI is for it to align information with care in real time, so it moves as decisions are made, not after the fact. When that happens, work across the care team becomes immediately actionable. Nurses can act on clear plans without waiting for note finalization, while care managers and behavioral health teams receive structured, usable context in real time instead of reconstructing it later. Physicians move from reconstructing visits after clinic to verifying decisions as they are made, reducing the need to revisit and rebuild work after the fact. In this model, documentation no longer trails care, it supports it.
Realizing this shift requires rethinking how AI fits into the full clinical environment. The next phase of clinical AI will be defined by how deeply it integrates into real workflows, from intake and nursing documentation to orders, care coordination, and structured data movement. Designing for this reality requires close partnership with clinicians and care teams, grounded in how care is actually delivered. AI must be architected with team-based care in mind and built for real-world deployment where trust, reliability, and adaptability matter as much as speed. Only then can it move from a point solution to infrastructure that supports the full care team.
Workflow redesign is an operational imperative.
Health care complexity is not easing. Patient needs are expanding, workforce shortages persist, and documentation demands continue to grow. This is an operational challenge, not an individual one. If we continue to frame burnout as an individual problem, we will keep reaching for individual solutions and seeing limited results. Well-being improves when workflows match the pace and complexity of real care delivery, and when each role is supported with tools aligned to how they work. Ambient AI is often positioned as a way to give time back to physicians. But if it only optimizes documentation for one role, it risks reinforcing the same fragmented system that created the problem in the first place.
Matt Sakumoto is a physician executive.
